{"schema_version":"onlylabs.public_analysis_evidence.v1","title":"Upstage (Solar) analysis evidence pack","description":"Public onlylabs evidence pack for cited agent analysis: captured pages, ranked public signals, and stored web-search provenance used by the background analysis workflow.","url":"https://onlylabs.fyi/analysis/upstage","json_url":"https://onlylabs.fyi/analysis/upstage/evidence.json","generated_at":"2026-06-28T02:19:03.549Z","org":{"slug":"upstage","name":"Upstage (Solar)","category":"neolab","category_label":"Neolab","dossier_url":"https://onlylabs.fyi/labs/upstage"},"analysis":{"url":"https://onlylabs.fyi/analysis/upstage","json_url":"https://onlylabs.fyi/analysis/upstage/analysis.json","generated_at":"2026-06-27T19:37:35.03+00:00"},"workflow":{"version":"onlylabs-deepagents-analysis-v3","provider":"deepseek","model":"deepseek-v4-pro","agent":"deepagents","public_pack_mode":"local-pages-and-events","live_web_fetches":false,"note":"Public evidence exports do not trigger live Exa calls; stored Exa provenance is included when analysis metadata contains it."},"stats":{"pages":28,"events":140,"web":0,"evidence":88,"signal_desks":{"hiring":0,"forks":11,"releases":23,"talking":12,"repos":14},"data_radar_lanes":null,"data_radar_matches":null,"stored_analysis_evidence":94,"stored_analysis_web":6,"stored_analysis_signal_desks":{"forks":11,"repos":14,"hiring":0,"talking":12,"releases":23},"stored_analysis_data_radar_lanes":null,"stored_analysis_data_radar_matches":null},"stored_web_provenance":{"queries":["\"Upstage (Solar)\" frontier AI lab recent model release research hiring GitHub Hugging Face","\"Upstage (Solar)\" AI lab what they are building talking about hiring releasing forking"],"request_ids":["585580a26c700f9539b3b81bd97c4919","868fff1ee4eb075caeabff992368dad2"],"skipped":null},"evidence":[{"ref":"P1","kind":"page","title":"The Modular Ai Tech Stack","date":"2026-06-27T00:01:40.53088+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/the-modular-ai-tech-stack","signal_url":null,"signal_json_url":null,"text":"The modular AI tech stack: How to build for a world where models change every year \n\nYour AI stack is already in motion. Make sure it holds up.\n\nMost insurance teams did not adopt AI through a strategy. They accumulated it. This guide shows what breaks as tools pile up and how to build a modular AI stack that stays accurate, traceable, and adaptable as models change.\n\nYou&#x27;re all set! Access your guide below.\nGet the guide \nOops! Something went wrong while submitting the form.\n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nIn most insurance organizations today, AI didn’t arrive through a strategy. It arrived through urgency.\nDoes this sound familiar? \nAn underwriting team starts using a general-purpose model to summarize submissions. \nA claims manager experiments with a tool to extract data from loss reports. \nA data engineer builds a small pipeline to structure PDFs for downstream analysis. \n\nNone of these efforts are coordinated, and none were designed to work together. They solve immediate problems in isolation, often without a shared understanding of how they fit into the broader system.\nFor a while, this works. Reviews move faster, teams feel more productive, and leadership sees early signals that AI is delivering value.\nThen, the cracks start to show.\nThis blog explores what’s actually breaking in today’s AI workflows, and how to build a modular AI stack that can keep up with ongoing model change .\nWhat breaks when AI is layered onto existing workflows\nConsider a mid-sized carrier processing commercial property submissions. A broker sends in a 300-page submission with schedules of values, loss runs, inspection reports, and supplemental documents. \nHistorically, reviewing a file like this could take days. Now, underwriters rely on a mix of tools to speed things up.\nOne underwriter pastes sections into a chatbot to summarize risk details. Another uses a separate tool to extract tables from PDFs. A third relies on an internal script built by IT to capture key fields. Each approach works in isolation, but none"},{"ref":"P2","kind":"page","title":"Askup Use Case Learn English","date":"2026-06-27T00:01:40.279446+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/ko/askup-use-case-learn-english","signal_url":null,"signal_json_url":null,"text":"AskUp으로 영어 공부하기 [똑똑한 AskUp 활용법 Vol. 4] \n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n비즈니스의 미래를 선도하는 인공지는 솔루션\n\n비즈니스 고민을 해결할 생성형 AI 모델이 필요하신가요?\n\n업스테이지의 AI전문가와 함께 시작하세요!\n\n✕ \n\n다양한 산업의 혁신을 주도하고 있는 AI가 교육과 학습의 패러다임도 바꾸고 있습니다. 이제 AI와 1:1로 공부하는 것이 더 이상 낯설지 않은 시대인데요. 영어라는 주제를 떠올렸을 때 누구나 한 번쯤은 해보았을법한 회화, 작문, 문법, 어휘 공부에 대한 고민들을 챗 GPT만 제대로 활용한다면 원어민 튜터와 공부하듯이 쉽고 재미있게 풀어볼 수 있습니다.\nLLM(대규모 언어 모델)은 사용자와 자연스러운 대화를 나누는 것에 최적화되어 있어 개인화된 피드백이 가능해 맞춤형 학습 을 이어갈 수 있는 것이 특징인데요. 가장 쉽고 빠르게 챗GPT를 만나볼 수 있는 카카오톡 속 AI 챗봇, ‘AskUp’과 함께 영어를 공부할 수 있는 무궁무진한 방법을 소개합니다!\n‍\nAskUp 영어 대화 모드 \n\n카카오톡에서 ‘AskUp’을 플러스 친구 로 추가하셨다면 하단의 “영어 대화 시작” 버튼을 눌러 보세요. 일반 대화 모드와 달리 AskUp이 영어로만 답변을 시작하게 됩니다. 이 영어 대화 모드의 특징은 바로 업스테이지가 LLaMA-2를 기반으로 파인튜닝하여 자체 개발한 생성 AI 모델 ‘SOLAR’를 통해 답변이 이루어진다는 점인데요.\n업스테이지는 지난 8월, 세계 최대 오픈소스 AI 플랫폼인 허깅페이스에서 챗GPT의 기반인 GPT-3.5의 성능을 뛰어넘고, Open LLM Leaderboard 1위를 기록하며 놀라운 기술력을 자랑한 바 있습니다. AskUp에서 새롭게 선보이는 영어 대화 모드를 통해 업스테이지의 LLM 모델을 체험해 보고, 일상 속에서 재미있게 영어 공부를 시작해볼까요?\n‍\n<일상 회화> \n스몰톡\n\nAskUp과의 영어 공부 방법 첫 번째, 바로 매일 한 두 가지의 주제로 일상적인 대화를 주고받는 연습 을 하는 것입니다. 많은 분들이 영어로 대화를 해야 한다고 생각하면 머릿속에 하고 싶은 말은 가득하지만, 어디서부터 어떻게 말해야 할지 몰라서 답답했던 경험이 있으실 텐데요. 이제 AskUp의 영어 대화 모드를 활용하여 간단하게라도 일상적이고 사소한 주제로 대화 나누는 것을 시작해 보세요.\n\n특히 모르는 사람과도 친근하게 대화하는 영어권 문화에서 스몰톡은 일상 회화의 중요한 역할을 차지하는 부분이기도 합니다. AskUp과 취미, 여행, 하루 일과, 좋아하는 음식, 관심사와 같은 소소한 주제에서부터 사회/과학 이슈에 대한 토론을 나누는 것까지 다양한 대화를 시도해보세요. 어떤 주제로 대화를 시작해야 할지 모르겠다면 AskUp에게 “Please give me a topic to talk about.”, “Please give me interesting topics to discuss.”와 같이 함께 이야기 나눌 수 있는 주제를 제안해달라고 요청 하는 것도 좋은 방법입니다.\n‍\n일상 영어 표현 알아보기\n\n‍\n또한 일상에서 자주 쓰이는 표현들도 알아볼 수 있습니다. 같은 말을 다르게 표현하는 방법이나 좀 더 자연스러운 구어체 표현 을 습득하는 데에 AskUp을 활용하여 실제 회화에 유용하게 적용해 보세요.\n‍\n<영단어 공부> \n특정 단어, 구동사가 들어간 예문 만들기\n\n‍\n언어를 구사하기 위해서는 무엇보다 어휘를 잘 아는 것도 중요합니다. 특정 단어나 구동사가 들어간 예문을 생성하여 읽어보면 실제로 어떻게 쓰이는지 더 잘 이해할 수 있는데요. 문장, 단락, 짧은 이야기 등 원하는 형식으로 특정 단어가 들어간 글을 생성할 수 있으며 단어를 여러개 포함하는 것도 조건으로 지정 할 수 있으니 다양한 예문을 통해 어휘를 학습해 보세요.\n‍\n퀴즈 만들기\n\n영어 단어 공부 하면 가장 먼저 떠오르는 학습 활동 중 하나가 바로 암기한 내용을 퀴즈로 만들어 복습하는 것이죠. AskUp에게 특정 단어나 주제를 먼저 주고, 그것을 활용하여 퀴즈를 만들어 달라고 요청하는 것도 방법입니다. 문장의 빈 칸을 채우는 형식, 반대어, 유의어, 뜻 풀이, 시제 변형 등 다양한 방식으로 퀴즈를 생성해 영어 단어를 복습할 수 있습니다.\n‍\n<영작문> \n문법 교정\n\n‍\nAskUp과 함께 영어 회화와 단어까지 공부했다면 작문까지 이어서 도전해 볼까요? 영어"},{"ref":"P3","kind":"page","title":"Ocr Api Free Trial Event","date":"2026-06-27T00:01:40.073679+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/ocr-api-free-trial-event","signal_url":null,"signal_json_url":null,"text":"Experience the amazing performance of OCR technology at the Upstage OCR API Launch Event \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nIntroduction \nMany companies are considering adopting OCR technology to improve work productivity. However, the reality is that it is difficult for many companies to easily apply it due to the resources required to develop and implement OCR technology, including high cost and issues with accuracy and performance.\nTo address these issues, Upstage plans to launch the OCR API service. When OCR API is applied, companies will not only be able to easily utilize OCR technology at a lower cost, but also attain focus for the essence of work as well as enhance work productivity.\n\nThree benefits of using OCR API \nCompared to on-premise OCRs, the advantages of the OCR API can be summarized in the following three ways:\n1. Since the OCR API is cloud-based, it provides higher scalability and flexibility .\n2. On-prem can meet more stringent requirements for security and privacy .\n3. OCR APIs allow users to easily start at no additional cost , while On-prem requires a higher initial investment.\n\nUpstage OCR API usage cases \nUpstage OCR API services can be used in a wide range of industries.\nResume Review : Companies that need to process numerous resumes in recruitment processes can utilize the Upstage OCR API service to automatically extract resume content, then review and analyze them with ease.\nReceipt Expense Processing : Businesses process a lot of expense receipts. Using the Upstage OCR API service, you can automatically extract the contents of the receipts and simplify expense processing and accounting tasks.\nImage Translation : The Upstage OCR API service provides a function that recognizes text in an image and translates it to a desired language. With this function, companies can quickly and accurately translate image documents written in various languages.\nInsurance Claims : In cases involving insurance claims, the Upstage OCR API service can automatically extract content from an insurance"},{"ref":"P4","kind":"page","title":"Document Parse Enhanced","date":"2026-06-27T00:01:40.062203+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/document-parse-enhanced","signal_url":null,"signal_json_url":null,"text":"Introducing Document Parse: Enhanced mode \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nUpstage is introducing Document Parse Enhanced, a new processing mode in Document Parse that extends standard parsing with direct understanding of visual document elements. This enables reliable extraction of complex tables, checkboxes, charts, and diagrams that previous text-based approaches cannot handle consistently.\nBest-in-class intelligence for complex document elements\nMost document parsing systems focus primarily on structured text, which creates clear limitations when documents rely on visual structure rather than explicit formatting.\nDocument Parse Enhanced mode leverages VLMs to remove these limitations and accurately understand:\n\nComplex tables — including multi-line cells, line-less tables, and multi-page tables\nCharts — converted into structured data and natural-language explanations\nImages and diagrams — summarized into concise, machine-readable descriptions\nCheckboxes — reliably detected with checked / unchecked status\n\nThese capabilities are available as part of Document Parse, allowing teams to work with irregular and visually complex documents in a clean, machine-ready form. See how complex tables and charts are parsed in the Document Parse Playground . \nHow Document Parse Enhanced mode works\nDocument Parse Enhanced mode builds on two core strengths of Document Parse—high OCR accuracy and reliable visual grounding—and extends them by applying vision language model to complex visual elements.\nThis allows Enhanced mode to understand how visual components relate to each other within the document layout, rather than treating them as isolated regions.\nAs a result, the system can:\nRecognize complex and line-less tables across pages\nConvert charts into structured values along with a narrative explanation\nSummarize images and diagrams so LLMs and downstream systems can understand their meaning\nDetect checkboxes and accurately recognize their checked or unchecked state\n\nPerformance benchmark: accuracy where document w"},{"ref":"P5","kind":"page","title":"Solar Pro 2","date":"2026-06-27T00:01:40.007824+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/solar-pro-2","signal_url":null,"signal_json_url":null,"text":"Introducing Solar Pro 2 \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nUpstage today officially released Solar Pro 2, its next-generation large language model (LLM) designed to perform complex reasoning and tool-integrated task execution—all with just 31 billion parameters.\nSolar Pro 2 builds on the foundation of its predecessor with a significant performance upgrade, increasing its parameter size from 22B to 31B while remaining in the small language model (sLLM) category. The model introduces a hybrid architecture with two user-selectable modes: “Chat Mode” for fast responses and “Reasoning Mode” for structured, multi-step logical thinking. Its Chain-of-Thought (CoT) reasoning approach significantly improves performance in advanced tasks like math, coding, and logic-heavy workflows.\n“Solar Pro 2 represents a new generation of AI agents that don’t just talk but think, reason, and act,” said Sung Kim, CEO and co-founder of Upstage. “Our mission is to fundamentally reshape how work gets done with AI, and Solar Pro 2 marks a key milestone in advancing that future of work.”\nIn benchmark tests such as MMLU-Pro, Math500, AIME, and SWE-Bench, Solar Pro 2 delivered results comparable to leading frontier-scale models including OpenAI’s GPT-4o, DeepSeek’s R1, Mistral’s Small 3.2, and Alibaba’s Qwen 3—despite being less than half their size. This performance validates the AI startup&#x27;s ability to develop globally competitive LLMs based entirely on homegrown technology.\nSolar Pro 2 also showcases remarkable strength in Korean language understanding. Across key benchmarks like Arena-Hard-Auto, Hae-Rae, and Ko-MMLU, the model matched or outperformed language models by tech giants. It demonstrates deep capabilities in vocabulary, context comprehension, and cultural nuance, while also excelling in high-stakes domains such as finance, medicine, and law.\nIn addition to its reasoning capabilities, Solar Pro 2 has evolved into a fully agentic LLM, capable of executing multi-step tasks by interacting with external tools. Given a pr"},{"ref":"P6","kind":"page","title":"Startup Upstage Branding","date":"2026-06-27T00:01:39.919725+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/startup-upstage-branding","signal_url":null,"signal_json_url":null,"text":"Upstage brand system building Part 1: Why Startups Need Branding \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n“It’s a startup, will we need to invest resources in branding?” 👀\n\n“Isn’t branding only for major corporations?” 🤔 \nMost startups think of these questions when they think of branding. For small startups and recently-formed businesses, branding is difficult to prioritize due to the time required and financials to be considered. On the other hand, as market competition intensifies, many companies often try new approaches while sparing no investments to create a more competitive brand.\nWhat exactly is a brand, and why do so many companies strive for branding? Is branding necessary for startups as well? \nThe early-stage startup Upstage, which currently prepares to launch a total AI solution encompassing all of the systems necessary to introduce AI technology within the year, has also shared its in-depth thoughts on this topic. During this process, Upstage summarized why branding is necessary and organized a unique brand system for itself.\nThrough this series, we review the reason startups need branding (Part 1) and provide insight on the Upstage brand system and its establishment process (Part 2).\nBrand essentials \nIf you take a look at domestic and international companies, there are those with very distinct colors. The images that come to mind when we think of companies such as Nike, Google, Apple, and Kakao are the result of their consistent and dedicated branding.\nBeyond global factors such as globally successful services, friendly customer service, exceptional promotional skills, and capable staff, companies still require a “clear brand identity” to develop their unique color. A brand’s identity has the power to attract people, just as we are captivated by another person’s unique personality, scent, style, and way of thinking.\nSo, why is a well-crafted branding system necessary?\n\nWhy startups need a branding system \nThere are five main reasons for why a brand system is needed:\n1. Consistency : Branding "},{"ref":"P7","kind":"page","title":"Underwriting Reinvention Diagnose And Elevate","date":"2026-06-27T00:01:39.91557+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/underwriting-reinvention-diagnose-and-elevate","signal_url":null,"signal_json_url":null,"text":"The 90 Path to Underwriting Reinvention: Phase 1 - Diagnose and Elevate \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nWhen control starts to look like progress\nIn insurance, control has always been a mark of discipline. Leaders are taught that multiple reviews, detailed audits, and strict sign-offs protect accuracy and uphold reputation. But over time, those same instincts have turned into barriers. The very systems designed to prevent mistakes now make it harder to move forward.\nAt Upstage, we see this dynamic almost everywhere we go. A carrier wants to modernize its underwriting operation, so leadership approves a series of new systems, each meant to make work smoother. A new intake portal here, an underwriting dashboard there, an additional reporting layer to “increase visibility.” On paper, it looks like progress. In practice, teams are juggling five logins, three approval steps, and two different versions of the same submission record. What was meant to simplify has become another form of control.\nThat’s why Phase One of our 90-day Modernization Plan exists.\nPhase One is the foundation of Upstage’s modernization model. It’s where we help carriers and MGAs step back before they step forward. The goal is to understand how work actually happens: how a submission travels, who touches it, and how many systems it moves through before it’s bound or declined. Together, we map those realities, identify what adds value and what simply adds friction, and decide where automation should step in and where human expertise should remain.\nPhase One isn’t about installing software. It’s about revealing the patterns that make software necessary. Once those patterns are visible, our AI Space platform helps organizations see what automation could look like in real terms. Specifically, we help reveal what it would mean for every submission, every touchpoint, and every workflow that currently depends on manual intervention.\nHow oversight becomes drag\n\nInsurance organizations are built on rigor, and that rigor is a strength. But whe"},{"ref":"P8","kind":"page","title":"Buy To Build How Upstages Modernization Playbook Empowers Underwriting Transformation","date":"2026-06-27T00:01:39.707801+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/buy-to-build-how-upstages-modernization-playbook-empowers-underwriting-transformation","signal_url":null,"signal_json_url":null,"text":"Buy to build: how Upstage’s modernization playbook empowers underwriting transformation \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nModernization is a priority for every insurer, yet the first step can be difficult to define. At Upstage, we’ve seen how that initial decision to build technology in-house or purchase a platform often determines whether transformation gains momentum or loses it. We believe in a smarter path: Buy to Build. By helping organizations invest in a modern, integration-ready platform and then building the transformation around it, we create a foundation for progress that lasts.\nTransformation doesn’t start with technology \nEvery modernization story begins before the first integration meeting. Phase One is about creating both technical and human clarity. It’s where organizations decide not only what they want to change but how they want to get there.\nAt Upstage, we start by listening to underwriters, operations leaders, and IT teams who live the daily friction of outdated workflows. We hear about manual document sorting, re-keying submission data, and the constant hunt for missing attachments. These are the pain points that prevent underwriters from spending time where it matters most: analyzing risk and serving clients. Phase One of our 90-day modernization plan is designed to uncover exactly where those inefficiencies live and how Agentic Information Extract can begin resolving them on day one.\nWhen legacy systems become liabilities \n\nBefore new technology can create progress, you have to see where the old processes are holding you back. We help teams map every step of their workflow, pinpointing the moments where human effort is spent verifying what a system could already know. This process isn’t about blame. It’s about visibility.\nIn underwriting, the patterns are familiar: hours spent verifying certificate details, reconciling COPE information, or retyping submission data between systems. With Agentic Information Extract, that manual load begins to disappear. The platform’s document-int"},{"ref":"P9","kind":"page","title":"Ai Tools For Business Work Productivity","date":"2026-06-27T00:01:39.687388+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/ko/ai-tools-for-business-work-productivity","signal_url":null,"signal_json_url":null,"text":"직장인의 업무 효율을 높여줄 AI 툴 \n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n비즈니스의 미래를 선도하는 인공지는 솔루션\n\n비즈니스 고민을 해결할 생성형 AI 모델이 필요하신가요?\n\n업스테이지의 AI전문가와 함께 시작하세요!\n\n✕ \n\n챗GPT의 등장으로 우리 일상의 많은 부분이 달라지고 있습니다. 특히 누구나 손쉽게 AI를 다룰 수 있는 선택지가 다양해지면서 업무에 활용하는 경우도 늘어나고 있는데요. ‘한겨레’에서 온라인 설문 플랫폼 ‘나우앤서베이’와 함께 지난 5일 직장인 500명을 대상으로 한 설문조사 결과에 따르면, 생성 인공지능이 업무에 도움이 된다는 응답이 전체의 80.6%(매우 도움 20.8%, 조금 도움 59.8%)에 이를 정도로 많은 분들이 AI의 유용함을 체감하고 있었습니다. AI 활용이 곧 경쟁력이 되는 시대에서 직장인의 업무 효율을 끌어올려 줄 수 있는 AI 툴을 살펴봅니다. \n‍\n올인원 작업공간, ‘노션 AI’ \n\n노션 AI 실행 화면 ‍\nNotion (노션)은 문서, 데이터베이스, 프로젝트 관리, 공개 웹사이트 등의 기능을 하나의 플랫폼에서 이용할 수 있도록 만든 올인원 생산성 툴로, 기업 내 협업툴로 활용되는 서비스이기도 합니다. 지난 2월, 본격 출시된 노션 AI는 OpenAI의 GPT-3 모델을 탑재하여 “아이디어 브레인스토밍 / 보도자료·블로그·SNS 게시물 등을 주제로 글쓰기 / 회의 아젠다 생성” 등 14가지 기능 을 제공하고 있습니다.\n\n아이디어 브레인스토밍에 ‘Digital transformation’을 입력하였을 때 ‍\n현재의 노션 AI는 생성 인공지능의 특성을 살려 메모와 문서 기능에 특화 되어 있습니다. 주요 기능을 살펴보아도 아이데이션이나 업무용 글쓰기에 도움을 받을 수 있다는 것이 한 눈에 들어오는데요. 위의 이미지처럼 아이디어 브레인스토밍 기능을 실행하고 사업이나 회의 등을 위해 고민이 필요한 주제를 입력한다면 AI가 자동으로 관련된 아이디어들을 제시해 줍니다.\n\n노션 AI로 새로운 초안 작성하기 (출처: 노션 도움말 센터 ) ‍\n글의 개요나 SNS 게시물 등의 초안을 작성하고 싶을 때는 메뉴에서 ‘초안 작성’ 옵션을 선택하고 프롬프트를 작성하면 페이지에서 AI가 작성한 내용을 확인할 수 있습니다. 아직 복잡한 글을 작성하고자 할 땐 사람의 손길을 한 번 더 거치는 것이 필요해 보이지만, 간단한 업무 고민을 덜어주는 데에는 손색없으니 14가지의 텍스트 생성 기능을 하나의 플랫폼에서 편리하게 이용하고 싶은 분들께 적합할 것 같습니다.\n‍\n내 손 안의 AI, ‘AskUp’ \n\n텍스트 생성 기능을 포함한 AI의 장점을 데스크탑과 모바일을 넘나들며 누리고 싶으신 분들은 ‘AskUp’을 주목해 주세요! 🔗 AskUp 은 인공지능 스타트업 업스테이지가 카카오톡에 론칭한 인공지능 챗봇으로, &#x27;챗GPT’에 업스테이지의 OCR 기술을 결합하여 사용자가 문서의 사진을 찍거나 전송하면 그 내용을 읽고 이해하고 답변할 수 있는 &#x27;눈달린 챗GPT’ 로 재탄생한 서비스입니다.\n\n사용자 편의를 위한 서비스 고도화를 지속하며 2021년까지의 정보만 학습한 챗GPT의 한계를 뛰어넘을 수 있는 ‘? 검색&#x27;(물음표 검색) 기능 과 자사 기술로 파인튜닝(finetuning) 된 이미지 생성 인공지능 모델 ‘업스케치(Upsketch)’를 통한 이미지 생성 기능 도 추가되어 AskUp과의 카카오톡 대화에서 모두 활용할 수 있게 되었습니다.\n\n업무에 쓸 수 있는 기능도 다양합니다. 비즈니스/영문 이메일 작성, 번역, 광고 문구 아이데이션, 링크 내용 요약, 궁금증 해결 등 사용자의 상황 및 목적에 따라 무궁무진하게 활용 가능합니다. 유료로 서비스 중인 타 챗GPT 기반 서비스와 달리 AskUp은 현재 무료로 매일 100건의 대화를 나눌 수 있어, 생성 AI를 경험하고 적극 활용해 보고 싶은 모든 분들께 좋은 선택지입니다.\n또한, AskUp의 B2B 버전 서비스로 ‘AskUp Biz’가 공개되기도 했는데요. 사내의 다양한 문서를 읽고 Chat AI로 정보를 얻을 수 있는 ‘AskUp Doc’, 홈페이지 정보를 읽고 홈페이지 방문자들에게 Chat AI를 통해 정보를 제공하는 ‘AskUp Web’, 업무용 툴인 슬랙에서 활용할 수 있는 ‘AskUp Slack’의 세 가지로 구성되어 있습니다. 비즈니스 환경에서 다양한 업무 과정을 혁신적으로"},{"ref":"P10","kind":"page","title":"Solar Pro Aws","date":"2026-06-27T00:01:39.565831+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/solar-pro-aws","signal_url":null,"signal_json_url":null,"text":"Upstage Releases Next-Generation “Solar Pro” Generative AI LLM on AWS \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n‍ SAN JOSE, Calif., Dec. 4, 2024 – Leading South Korean artificial intelligence (AI) startup Upstage today announced that it has launched its next-generation large language model (LLM) Solar Pro, on Amazon Web Services (AWS). Available now on Amazon Bedrock Marketplace, Amazon SageMaker JumpStart and AWS Marketplace, Solar Pro can be easily customized and fine-tuned across a range of industries like healthcare, finance, and legal services.\nAt 22B parameters, Solar Pro, a larger model than Upstage’s previous 10.7B parameter model Solar Mini, shows 50% improvement in performance across key benchmarks at lower cost. In its September preview release, Solar Pro topped HuggingFace’s Open LLM Leaderboard for models under 70B parameters, excelled on the EQ Bench Leaderboard for emotional intelligence, and outperformed all other open-source models from major tech companies on Predibase&#x27;s Fine-Tuning Leaderboard.\nSolar Pro was trained on Amazon SageMaker, a fully managed machine learning (ML) service, that reduced training time significantly through advanced data pre-processing (steps taken to prepare and clean input data before it is used for training) and continued pre-training techniques. Data from the 1 Trillion Token Club, an Upstage-founded alliance for Korean-specialized LLMs, was used for training. Consisting of a variety of copyright-free English and Korean training data from texts, books, news articles, reports, the data evolved Solar Pro’s understanding of cultural nuances while ensuring higher accuracy in responses, mitigating “AI hallucinations” that generate false or inappropriate answers.\nThe 22B model is optimized for single-GPU deployment, utilizing Upstage&#x27;s proprietary Depth-Up Scaling (DUS) method—a pre-training technique that ensures a compact model size without sacrificing performance. This method involves carefully redesigning their existing AI architecture to maintain high per"},{"ref":"P11","kind":"page","title":"When Ontology Moves Faster Than It","date":"2026-06-27T00:01:39.562318+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/when-ontology-moves-faster-than-it","signal_url":null,"signal_json_url":null,"text":"When Ontology Moves Faster Than IT — How Upstage Keeps You Ahead \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n‍ Your Ontology Changes Every Quarter. Your Automation Shouldn&#x27;t. \nLast quarter, your insurance forms had 47 fields. This quarter? 63. Next quarter? New products, new regulations, new data points. Ontology changes faster than any IT sprint. \nHere&#x27;s the problem : Most document automation treats data structures as permanent. Build it once, use it forever. But in the real world:\nNew insurance riders add fields overnight\nRegulations split one field into three\nMergers introduce different terminology\nProduct launches require new data points\n\nTraditional automation breaks every time. IT rebuilds the system. Operations waits. Documents pile up. Manual processing fills the gap.\nCan your automation keep up? \nThe Upstage Difference: Built for Change\nWe designed an extraction system that adapts to your data structure instantly, whether it&#x27;s today&#x27;s, tomorrow&#x27;s, or the one after your next merger.\nExtract Any Schema with Information Extract\n\nSchema as configuration, not trained model. Add fields, update descriptions, deploy in minutes. When your insurance forms go from 47 to 63 fields, you update the schema. No retraining. No IT sprints.\nQuick deployment: \n3-minute auto schema generation. Upload 10 documents, AI suggests fields, review, deploy. 80% accuracy out of the box\nFew-shot examples for terminology. Add 2-3 examples for new terms. +6.65% accuracy improvement in internal tests\n\nBuilt-in validation: \nConfidence scores: Low-confidence values flagged for review. Focus effort where it matters\nSource tracing: Check any value&#x27;s location (page, coordinates) in the original document\nAuto-improvement : Errors feed back to update schema and propose improvements\n\nPrerequisite: Any document format (PDF, scans, forms) is normalized through Document Parse, so format changes never break extraction. \nQ: What if 50 new fields are added overnight?A : Update schema, validate sample batch, deploy. It ju"},{"ref":"P12","kind":"page","title":"Ai Ethics","date":"2026-06-27T00:01:39.501065+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/ai-ethics","signal_url":null,"signal_json_url":null,"text":"The need for AI ethics and corporate efforts \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nArtificial Intelligence (AI) unquestionably offers abundant benefits to our lives and society. It greatly contributes to improving convenience across a wide range of sectors, including finance, healthcare, education, and logistics. However, it is not only convenience that the advancement of AI brings; it also raises concerns and anxieties. These concerns stem from worries that AI may infringe on human rights and freedom, weaken responsibility and creativity, and exacerbate social inequality.\nTherefore, there is a growing voice advocating for AI ethics and reliability, emphasizing the need for AI to be designed and operated in a way that respects and upholds human dignity and the common good. On January 1st, the first-ever AI Safety Summit was held in the United Kingdom, marking a global effort to ensure reliable AI. Let&#x27;s delve into how we can ensure our well-being and safeguard our lives in a world with AI and how businesses are seeking solutions to these challenges together.\nWhat is AI Ethics? \nAI ethics refers to a set of principles, values, and guidelines that advise the ethical and responsible use of artificial intelligence in its design and outcomes. According to a government publication titled \"Artificial Intelligence Technology Outlook and Innovation Policy Directions,\" AI ethics is defined as \" universal social norms and relevant technologies that stakeholders in AI should abide by\" in a contextual definition.\nThere are many ethical considerations in the use and development of AI, with particular importance placed on awareness and response to inherent cognitive biases in the data used for training. AI learns from various data sources, and these data often contain biases inherent in human society. Therefore, it is crucial to construct experiments and algorithms that minimize biases and enhance fairness and diversity. \nRenowned scholar in the field of AI ethics, Professor Sandra Wachter of the Oxford Internet Ins"},{"ref":"P13","kind":"page","title":"Upstage Ai Talk Show Recap","date":"2026-06-27T00:01:39.343324+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/upstage-ai-talk-show-recap","signal_url":null,"signal_json_url":null,"text":"Recap for Upstage AI Talk Shows in 2024 \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nAt Upstage, we are dedicated to fostering a vibrant ecosystem for large language models (LLMs). We have pursued it every step of the way to shape the future of artificial general intelligence for work . From June to September, we had the pleasure of hosting two sessions of the \"Upstage AI Talk\" in Seoul, in partnership with SBVA , a prominent global venture capital firm.\nThese two sessions brought together over 300 attendees from AI startups and developers. From thought-provoking discussions to panel talks with Q&A sessions, lively discussions took place between the keynote speakers and attendees. Through these, speakers and attendees were able to share valuable insights into LLMs and delve into the future of artificial intelligence.\nKey Highlights \n\nSession 1: June 24, 2024 \nKeynote speakers \nRichard Socher : Founder and CEO of you.com , the first chat-search assistant.\nCindy Jin:  Partner at SBVA , one of the most prestigious venture capitals in Asia.\nSung Kim : Co-Founder and CEO of Upstage.\n\nTakeaways ‍ \nRichard Socher discussed the impact of AI development on the industry, emphasizing that generative AI will enable the realization of many previously impossible tasks. He addressed on \" You.com - Personalized AI Chat\" highlighted the incredible potential of AI-powered search engine to revolutionize the way we interact with technology. He shared insights into the latest advancements in natural language processing and machine learning, showcasing how AI can be leveraged to provide personalized and engaging user experiences.\nSung Kim discussed the societal transformations expected from the fast-paced evolution of AI. His keynote on \"Solar: The Best Fine-Tunable LLM\" delved into Upstage&#x27;s groundbreaking work on large language models (LLMs) and the capabilities of Solar Mini to solve business problems with super-human level intelligence, presenting various real-world industry use cases.\n\nSession 2: Sep 11, 2024 \nKeynote Speake"},{"ref":"P14","kind":"page","title":"Ocr Information Extraction","date":"2026-06-27T00:01:39.252495+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/ocr-information-extraction","signal_url":null,"signal_json_url":null,"text":"Technology for digital assetization 101 \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n(This blog is Part 2 of the Digitize anything! series.)\nAs mentioned in the previous blog, digital assetization must precede digital transformation . However, depending on the purpose of this assetization, the technology used will vary. So which technology should I use? In this article, we will explain which technologies to adopt based on purpose and data.\n\nOCR: A technology that reads all letters of a document \nIf our organization’s data is in the form of an image or document file, and we want to find and read every letter of the file, we can use OCR—Optical Character Recognition.\nWhat input does OCR take? Document files such as PNG, JPG, PDF, etc.\nWhat output does OCR return? Characters and character location information.\nHow does OCR work? OCR consists of two models. First, there is a detector that finds letters from a given file, and a recognizer that deciphers which letters it finds.\n\nDetector → Recognizer In the case of the detector, the position of the letter is expressed as a quad (square of 4 points), polygon (contour expressed as 2N points), center point (one center point), etc. Upstage Document OCR detects using a rectangular method with four dots, as shown in the photo above.\nConversely, for the recognizer, character recognition is performed based on predefined target characters. Undefined characters are usually recognized as unknown symbols such as “�”. Currently, the characters targeted for recognition defined by Upstage Document OCR include:\n(1) Korean\n(2) English\n(3) Numbers\n(4) Chinese characters\n(5) Special characters\nWe are continuously updating this list according to customer requests.\nNowadays, both detectors and recognizers are developed as deep-learning based models and, depending on the scenario in which it is utilized, an end-to-end integrated model can be substituted for developing the detector and recognizer separately.\n\nWhat does a good OCR model constitute? \nTo adopt the best OCR model for your compan"},{"ref":"P15","kind":"page","title":"How To Use Ai Chatbot Askup 1","date":"2026-06-27T00:01:39.191535+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/how-to-use-ai-chatbot-askup-1","signal_url":null,"signal_json_url":null,"text":"AskUp official user guide Vol. 1 - adding channel friends \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nThe performance of &#x27;AskUp&#x27;, launched by Upstage on KakaoTalk using ChatGPT and OCR (Optical Character Recognition) technology, is extraordinary. The number of channel friends on KakaoTalk exceeded 100,000 within a week , and there are now about 300,000 people who have registered as AskUp&#x27;s channel friends.\nThrough the Upstage blog series, we would like to explain the correct use of AskUp alongside its various uses , which, once you are familiar, can be utilized like a mini-encyclopedia anytime, anywhere. If you are curious about AskUp, please visit the Upstage Blog!\nLearn about &#x27;AskUp&#x27; \n\nThe profile image of AskUp before its brand renewal was created by the image-creating technology “Generative AI”. A friendly AI chatbot that responds pleasantly to various questions became the official mascot.\nAskUp is an artificial intelligence chatbot service developed by Upstage. It was named by combining ‘Ask,’ to ask questions, and our AI company name ‘Upstage.’ In Korea, the alternative nickname ‘Asukup’ is easier to pronounce.\nAskUp began when Upstage linked the internal work tools ChatGPT and Slack together,  solving simple tasks and questions. The current AskUp was created by expanding the service to KakaoTalk in hopes that more people would know about and use the convenient technology of AI.\nThe main feature of AskUp is the so-called ‘Awakened ChatGPT,’ which combines Upstage&#x27;s OCR technology and ChatGPT so users can discover text in an image by taking or sending a picture.\n\nAskUp&#x27;s strengths include:\nOCR technology : AskUp accurately recognizes text regardless of font, background, etc. Through this, users can send text documents or even handwritten images to AskUp and receive answers along with translated content. In addition, when asked about various documents such as learning materials, business registration certificates, or contracts,  AskUp summarizes necessary information or pro"},{"ref":"P16","kind":"page","title":"Evalverse Llm Evaluation Opensource","date":"2026-06-27T00:01:39.140131+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/evalverse-llm-evaluation-opensource","signal_url":null,"signal_json_url":null,"text":"Evalverse: Revolutionizing large language model evaluation with a unified, user-friendly framework \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nIn the rapidly advancing field of artificial intelligence, evaluating Large Language Models (LLMs) is often a complex and disjointed task. Acknowledging the necessity for a more integrated method, Upstage proudly presents Evalverse, an innovative library designed to simplify and unify the evaluation process. This tool not only facilitates a more systematic assessment of LLMs but also makes cutting-edge evaluation techniques accessible to a broader audience, ensuring that advancements in AI are both inclusive and comprehensive.\n\nOverview of Evalverse What is Evalverse? \nEvalverse is a centralized platform designed to streamline the evaluation of LLMs by integrating a variety of evaluation methodologies. It incorporates well-known frameworks such as lm-evaluation-harness and FastChat as submodules. This architecture enables Evalverse to serve as both a unified and expandable library and simplifies the process of updating, ensuring the tool remains at the forefront of technological advancement.\nKey features \nUnified evaluation with submodules: Evalverse leverages Git submodules to integrate and manage external evaluation frameworks, such as lm-evaluation-harness and FastChat . This approach allows for the straightforward addition of new submodules, facilitating the support of a broader range of evaluation frameworks. Moreover, it enables the seamless incorporation of upstream changes, keeping Evalverse up-to-date in the dynamic landscape of LLM technology.\nNo-code evaluation request : Evalverse introduces a no-code evaluation feature, accessible through Slack interactions. Users simply initiate a request by typing Request! in a direct message or a designated Slack channel with an active Evalverse Slack bot. The bot then guides the user through selecting a model from the Huggingface hub or specifying a local model directory, culminating in the execution of the evaluation proc"},{"ref":"P17","kind":"page","title":"Solar Api Beta","date":"2026-06-27T00:01:38.99581+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/solar-api-beta","signal_url":null,"signal_json_url":null,"text":"Introducing Solar API Beta \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nSeoul, Feb. 22, 2024 – Upstage is thrilled to announce the beta release of Solar, its revolutionary Large Language Model (LLM). Until March 31st, developers and businesses can access Solar&#x27;s powerful capabilities through a free API, unlocking the potential to transform their operations.\nSolar is a pre-trained LLM from Upstage, which stunned the AI community in December 2023 by topping the Open LLM Leaderboard on Hugging Face. Remarkably, the model achieves performance comparable to GPT-3.5 while using fewer parameters and operating at 2.5 times faster speeds. Solar also outshined competitors like Llama2, Mistral 7B, Ko-Alpaca, and KULLM in a range of benchmarks, demonstrating the company&#x27;s expertise in downsizing LLMs without compromising performance.\nBeyond its impressive efficiency, Solar API Beta offers several compelling features:\nEnglish-Korean translation: This API-accessible version surpasses industry giants like GPT-4 and DeepL in translation quality, thanks to Upstage&#x27;s innovative context-aware approach. Solar&#x27;s ability to understand the broader conversation context, extending beyond individual sentences, leads to superior translation accuracy and nuance.\nSolar LLM Innovators Award: Upstage is fostering a vibrant community of innovators by launching the Solar LLM Innovators Award. Up to 10 companies or organizations will receive $200,000 worth of Solar API credits for showcasing unique and impactful business use cases powered by Solar.\n\nWith its exceptional performance, innovative features, and commitment to fostering creativity, Solar is poised to revolutionize the LLM landscape. Upstage invites developers and businesses to embrace the future of AI and experience the transformative power of Solar.\n\nRelated News\n\nBrowse all Articles\n\nMarch 19, 2026\n\nAMD and Upstage Expand Strategic Collaboration to Advance Sovereign AI Infrastructure in Korea\n\nDecember 17, 2025\n\n2025 in Review: South Korea’s Leading AI Innovator M"},{"ref":"P18","kind":"page","title":"Upstage Aws Sca Global","date":"2026-06-27T00:01:38.853498+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/upstage-aws-sca-global","signal_url":null,"signal_json_url":null,"text":"Upstage and AWS Join Forces to Drive AI Innovation in APAC and the United States \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nUpstage today announced a strategic collaboration with Amazon Web Services (AWS) to drive co-sales and go-to-market efforts while advancing the development of its Solar language models to broaden access to high-performance, cost-effective generative AI technology across the Asia-Pacific (APAC) region and the United States. In addition, Amazon will make a minority investment in Upstage, a milestone that underscores Upstage’s technical leadership and growing influence in the region.\nUnder this collaboration, Upstage will work with AWS as its preferred cloud provider to build, train, and deploy its future foundation models. The company will leverage AWS’s world-leading machine learning infrastructure, including Amazon SageMaker, AWS Trainium and Inferentia chips, to scale its large language model (LLM), Solar, and a suite of AI-powered document processing solutions.\n“Upstage and AWS have built a strong foundation through years of collaboration, and this agreement represents a new chapter in our shared vision to accelerate AI innovation globally,” said Sung Kim, CEO and co-founder of Upstage. “With AWS as our strategic partner, we’re poised to scale our impact and bring secure, intelligent, and high-performing AI solutions to more organizations in the public sector.”\nTogether, the companies aim to empower governments, public agencies, and regulated industries with best-in-class generative AI solutions, optimized for the performance, security, and the compliance needs of the public sector. As part of the agreement, Upstage and AWS will collaborate on go-to-market strategies and co-selling efforts across APAC and the US. \n“AWS is excited to enable Upstage to unleash a new wave of innovation”, said Jon Jones, VP and Global Head of Startups and VentureCapital at AWS. “From our fully-managed AI services to purpose-built chips for training and inferencing of foundation models, Upstage is leveraging "},{"ref":"P19","kind":"page","title":"2pyeon Api Ereokodeu Wanjeon Haebu","date":"2026-06-27T00:01:38.833274+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/ko/2pyeon-api-ereokodeu-wanjeon-haebu","signal_url":null,"signal_json_url":null,"text":"Solutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n비즈니스의 미래를 선도하는 인공지는 솔루션\n\n비즈니스 고민을 해결할 생성형 AI 모델이 필요하신가요?\n\n업스테이지의 AI전문가와 함께 시작하세요!\n\n✕ \n\nAPI를 사용하다 보면 가장 먼저 맞닥뜨리는 것이 에러 코드입니다.\n예상치 못한 응답을 받으면 당황하기 쉽지만, 사실 대부분은 원인을 확인하고 몇 가지 사항만 점검하면 빠르게 해결할 수 있습니다.\n이 글에서는 Upstage API에서 자주 발생하는 에러 코드를 정리하고, 각각의 원인과 해결 방법을 안내드립니다.\n\n🔑 인증 및 권한 관련\n401 Unauthorized\n원인 잘못된 API Key 또는 Authorization 헤더 오류\n\n확인 사항 API Key가 요청 헤더에 포함되어 있는지 확인\nAuthorization: Bearer API_KEY 형식이 맞는지 점검\n콘솔에서 해당 API Key가 삭제된 것은 아닌지 확인\n오타, 따옴표, 공백 등 단순 실수가 없는지 검토\n\n403 Forbidden\n원인 사용 가능한 크레딧 부족\n결제 실패\n내부 정책에 따른 특정 IP 차단\n\n해결 방법 콘솔에서 크레딧 잔액 확인\n등록된 결제 수단이 정상인지 확인\n문제가 지속되면 지원센터 로 문의\n\n📄 요청 형식 관련\n400 Bad Request\nSolar APIs JSON 구문 오류\n필수 키 누락\n키 이름이나 값 타입 불일치\n값 오류 또는 오타\n\nDocument AI APIs 파일 경로 오류\n지원하지 않는 파일 포맷 업로드\n50MB를 초과하는 파일\n\n404 Not Found\n원인 : 잘못된 URL Path\n해결 방법 : 요청 경로를 다시 확인\n\n405 Method Not Allowed\n원인 :  http method (GET, POST, 등) 중에 저희가 허용하지 않은 method 로 요청하는 경우에 발생\n해결 방법 : docs 에서 호출하려는 API 의 url 과 method 가 정확한지 확인 후, 맞는 url, method 로 요청\n\n⏱️ Rate Limit 및 과부하\n429 Too Many Requests\n원인 : 초당 요청 수(RPS) 또는 분당 토큰 수(TPM), 또는 분당 요청 수(RPM) 제한 초과\n해결 방법 요청 간격을 늘리거나 큐를 통해 순차 처리\nRate Limit 증가 신청 가능\n\n💥 서버/시스템 오류\n500 Internal Server Error\n원인 : 일시적인 서버 오류, Document Parse Async 처리 실패\n해결 방법 : 잠시 후 재시도 → 문제가 반복되면 상태 페이지 확인 후 지원센터 문의\n\n502 / 503 / 504\n원인 : 게이트웨이/서비스 가용성 문제\n해결 방법 일시적 장애일 가능성이 높음\n요청을 재시도하고, 동일 현상이 지속되면 지원센터로 문의\n\n🧩 Document Parse Async 특수 사례\n403 Client Error (download_url 만료) download_url은 발급 후 약 15분간만 유효\n원래 request_id 로 다시 조회하면 새 URL 발급 (추가 과금 없음)\n\n“Failed to process document” 파일명이 지나치게 긴 경우 (한글 ≤ 300자, 영문 ≤ 900자 권장)\n문서 구조나 형식에 따른 처리 제한\nTechnical support 신청 을 통해 지원팀에 request_id 와 샘플 문서를 전달하면 빠른 원인 파악을 위한 도움을 받을 수 있음\n\n요약\n400 : 요청 형식 오류 (JSON/파일 경로/50MB 제한)\n401 : API Key 또는 Authorization 헤더 문제\n403 : 크레딧 부족, 결제 실패, IP 차단\n404/405 : 잘못된 URL Path, http 대신 https 필요\n429 : Rate Limit 초과 → 요청 빈도 조정 및 상향 신청 가능\n500~504 : 서버/게이트웨이 오류 → 재시도 후 지속 시 지원센터 문의\nDocument Parse Async 특수 케이스 : download_url 만료, 파일명·문서 구조 문제\n\nQuick FAQ Recap\nQ. 400 Bad Request가 자주 발생하는 이유는 무엇인가요? \nA. 파일 경로·형식·크기(50MB 이하)를 점검하세요.\nQ. 401 Unauthorized는 어떻게 해결할 수 있나요? \nA. Authorization 헤더를 \"Bearer API_KEY\" 형태로 정확히 입력했는지 확인해 주세요.\nQ. 403 For"},{"ref":"P20","kind":"page","title":"2023 Ai Tech Trend Seargest","date":"2026-06-27T00:01:38.796614+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/ko/2023-ai-tech-trend-seargest","signal_url":null,"signal_json_url":null,"text":"[AI 트렌드] 매출 성장의 비결로 떠오른 검색· 추천 기술, Seargest \n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n비즈니스의 미래를 선도하는 인공지는 솔루션\n\n비즈니스 고민을 해결할 생성형 AI 모델이 필요하신가요?\n\n업스테이지의 AI전문가와 함께 시작하세요!\n\n✕ \n\n커머스/콘텐츠 플랫폼 필수 기술, Seargest(search+suggest) \n사용자의 데이터를 바탕으로 초개인화된 검색·추천 기능을 제공해주는 “서제스트(Seargest; search+suggest)” 를 알고 계신가요? Seargest는 AI로 검색하고 추천하는 기능을 합친 것으로, 비대면 문화가 확산되며 이미 우리의 일상 곳곳에 적용되고 있습니다.\n이제는 TV 본방 사수 시대를 넘어 언제 어디서나 자신이 원하는 시간에 취향에 맞는 콘텐츠를 볼 수 있게 해주는 유튜브 및 OTT 서비스와 비대면으로 손쉽게 활용하는 온라인 쇼핑 플랫폼에서도 Seargest는 고객의 서비스 만족도 및 기업 매출 증대에 가장 큰 영향을 미치는 필수 기술 로 자리매김 하고 있습니다.\n초개인화된 검색·추천 기술이 콘텐츠와 커머스 플랫폼에서 더욱 중요해지고 있는 이유는 크게 두 가지입니다.\n🔎 검색 기능은 온라인상의 정보 홍수 속에서도 사용자가 원하는 정보를 찾을 수 있도록 도움 \n📈 사용자의 취향에 맞는 콘텐츠나 상품을 추천하여 서비스 이용 시간 및 구매 전환율이 높아짐 \n\n그렇다면 실제 비즈니스에서 검색·추천 기술, Seargest는 어떻게 활용되고 있을까요? 다양한 기업의 사례를 살펴보겠습니다.\n‍\n콘텐츠-커머스 플랫폼의 Seargest 적용 사례 \n1. 아마존 \n\n상품 검색에 AI를 적용한 아마존 글로벌 쇼핑 플랫폼 서비스를 운영하는 아마존은 영어가 아닌 다른 언어의 감정어도 자동 번역하여 적합한 제품을 찾아 고객의 구매를 이끌어내고 있습니다. 이를테면 검색창에 “쌀쌀할때”를 입력했을때 따뜻하게 입을 수 있는 옷이 검색 결과로 보여지는 것입니다. 마치 우리가 쌀쌀한 날씨를 생각했을 때 긴팔 티셔츠나 외투를 떠올리는 것처럼 AI도 검색어를 똑같이 이해하는 것입니다. 이처럼 상품 검색에 AI를 적용하면 오타 수정과 번역은 물론, 자연어 처리를 기반으로 모호하고 복잡한 키워드도 의미 중심으로 가장 적합한 검색 결과를 도출 해낼 수 있습니다.\n또한 아마존이 만든 세계 최대 퍼블릭 클라우드 서비스인 AWS는 자사 플랫폼을 바탕으로 발전시킨 AI 기술을 기반으로 지능형 검색 서비스인 ‘아마존 켄드라’와 개인화 추천 서비스인 ‘아마존 퍼스널라이즈’를 출시하기도 했습니다.\n‍\n2. 유튜브 \n\n유튜브의 추천 AI 알고리즘 (유튜브로 음악을 즐겨듣는 사용자에게 추천된 콘텐츠 화면) ‍\n유튜브는 ‘AI 알고리즘’ 의 대명사가 될 정도로 AI 추천 기술 이 빛을 발하는 앱입니다. 와이즈앱의 조사 결과 유튜브는 한국인이 가장 오래 사용하는 앱으로 선정될 정도로 우리 생활과 뗄 수 없는 존재가 되었습니다. 이러한 유튜브의 승승장구는 사실 AI의 힘이라고 봐도 과언이 아닙니다.\n닐 모한 유튜브 최고상품담당자(CPO)는 지난해 뉴욕타임스 인터뷰에서 “유튜브에 AI 알고리즘을 도입한 이후 총 시청 시간이 20배 이상 증가했다” 라고 밝히기도 했습니다. 초개인화된 Seargest 기술로 사용자들의 취향에 맞는 콘텐츠를 추천해주다보니 자연스럽게 플랫폼 이용률이 극대화된 것입니다.\n‍\n3. 네이버 \n\n네이버의 에어서치 검색 기능 중 ‘스마트블록’ 적용 화면 ‍\n국내 포털 점유율 1위의 네이버 또한 검색·추천 기술 Seargest를 적극적으로 활용하고 있습니다. 네이버는 에어서치 검색 기능을 통해 스마트블록, 지식 인터렉티브, 옴니서치, 동영상 장면 탐색 기능, 웹 검색 결과를 고도화하고 있습니다. 특히 에어서치 검색 기능 중 ‘스마트블록’은 검색 사용자의 의도, 취향을 반영하여 사용자별 최적의 맞춤형 검색 결과를 스마트블록 단위로 세분화 해 보여주는 것입니다. AI가 사용자의 검색 의도를 파악해 초개인화된 콘텐츠를 노출함으로써 사용자의 경험이 향상되고 있는 것이죠.\n\n또한 네이버는 큐레이션 서비스를 강화하고자 자사의 AI 기반 상품 추천 기술을 활용해 사용자 개인의 관심사와 취향을 모은 AI 쇼핑 큐레이션 공간 ‘포유’ 탭 을 신설하기도 했습니다. 네이버쇼핑 내 약 10억개의 상품 데이터베이스(DB)중 자신의 관심사와 취향에 맞는 아이템을 실시간"},{"ref":"P21","kind":"page","title":"Solar Pro 2 Frontier","date":"2026-06-27T00:01:38.703118+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/solar-pro-2-frontier","signal_url":null,"signal_json_url":null,"text":"Solar Pro 2 Breaks Into Global Frontier AI \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nUpstage has earned global recognition with its latest proprietary large language model (LLM), Solar Pro 2, which was recently named one of the top 10 “Frontier Models” by leading AI benchmark organization Artificial Analysis. It is the only model developed in Korea to be included in the index, placing Upstage alongside industry leaders such as OpenAI, Google, Meta, and Anthropic.\nArtificial Analysis’ Intelligence Index is regarded as one of the most comprehensive public evaluations of language model performance. The index assesses over 200 models using seven key benchmarks—including reasoning, coding, general knowledge, and advanced math—to generate a composite score that reflects multiple dimensions of intelligence.\nSolar Pro 2 achieved a score of 58, outperforming widely used models such as GPT-4.1 (53), GPT-4o (41), and LLaMA 4 Maverick (51). It also surpassed notable newcomers such as Kimi-K2 (57.59) and DeepSeek V3 (53)—a significant achievement for a model with just 31 billion parameters. In contrast, many of the models it surpassed or matched operate at 100B+ scale, with the top-ranked Grok-4 reportedly reaching 1.7 trillion parameters.\nThe recognition highlights Solar Pro 2 as a rare example of a compact model delivering near-frontier performance. In addition to outperforming larger models in reasoning-heavy tasks, it also ranked highly in cost-efficiency metrics, making it a viable option for enterprise use where performance and cost must be carefully balanced.\nWhile GPT-4.1 remains a dominant force in real-world applications, especially across enterprise AI deployments, Solar Pro 2’s higher score suggests stronger capabilities in complex reasoning and knowledge synthesis—functions increasingly central to the next generation of AI systems.\nThe inclusion of Upstage in the “Frontier Model” category marks a milestone for the Korean AI ecosystem and reflects a growing diversification in global AI leadership. Upstage is no"},{"ref":"P22","kind":"page","title":"Solar Sendbird","date":"2026-06-27T00:01:38.52909+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/solar-sendbird","signal_url":null,"signal_json_url":null,"text":"Solar LLM Powers Sendbird&#x27;s No-Code AI Chatbot \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nSeoul, Mar. 6, 2024 – Upstage announced today the integration of Solar, the company’s advanced pre-trained large language model (LLM), into Sendbird&#x27;s no-code generative AI chatbot. This partnership aims to enable business owners and companies to easily deploy generative AI chatbots on websites and mobile apps, providing human-like services in customer support and engagement.\nSolar LLM boasts a compact yet high-performance design in various natural language processing (NLP) tasks, standing on par with industry titans. The 10.7B model made waves in the global AI stage last December, securing the top position across benchmark tests on the Hugging Face Open LLM Leaderboard and surpassing even larger models like Mixtral 8x7B.\nSendbird powers interactions between 300 million users globally every month with its enterprise chat interface, offering services to more than 4,000 enterprise companies. The company’s AI chatbot allows for the creation and deployment of AI chatbots specializing in customer service by providing just a few pieces of information, without needing a large engineering team.\nWith a convenient API access to Solar, AI chatbot users worldwide can leverage the power of the state-of-the-art LLM to generate highly accurate and personalized responses to their queries. Solar brings unmatched robustness and adaptability to enterprises across diverse industries, providing exceptional flexibility for customization in areas such as finance, commerce, and customer management.\nSung Kim, CEO of Upstage, added, \"We are thrilled to integrate Solar into Sendbird&#x27;s world-class chat solution, and further innovate the overall customer journey experience that Sendbird provides through Solar.\" \nJohn Kim, CEO and Co-founder of Sendbird, expressed, \"We anticipate offering solutions that fully leverage the potential of generative AI by combining Smart Assistant with Solar,\" emphasizing, \"It will become a game-changer for "},{"ref":"P23","kind":"page","title":"Kbank Mou","date":"2026-06-27T00:01:38.468815+00:00","date_source":null,"source_url":"https://www.upstage.ai/news/kbank-mou","signal_url":null,"signal_json_url":null,"text":"Upstage Forge Partnership to Drive AI Innovation in Finance \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nSeoul, Feb. 27, 2024 – Upstage signed a Memorandum of Understanding (MOU) with KT, kt cloud, and K Bank to collaborate on leveraging large language models (LLMs) in the financial sector.\nThis partnership aims to establish a customized LLM environment tailored for the financial sector. This paves the way for the secure deployment of cutting-edge generative AI solutions, addressing security concerns and the “AI hallucinations”, generation of false or misleading information by LLMs.\nUnder the MOU, Upstage will spearhead the technical department, utilizing its industry-leading LLM, \"Solar,\" and fine-tuning it with specialized financial data. KT and kt cloud will provide the GPUs and computing infrastructure to train the model. K Bank will contribute by identifying potential applications for AI integration within its various offerings and evaluating the effectiveness of the LLM implementation.\nSung Kim, CEO of Upstage, said, “We are thrilled to partner with KT, kt cloud, and K Bank to spearhead LLM innovation in the financial sector. We are confident that our world-class technology will empower various industries, starting with finance, to strengthen their AI competitiveness.\"\n\nRelated News\n\nBrowse all Articles\n\nMarch 19, 2026\n\nAMD and Upstage Expand Strategic Collaboration to Advance Sovereign AI Infrastructure in Korea\n\nDecember 17, 2025\n\n2025 in Review: South Korea’s Leading AI Innovator Marks Breakout Year After U.S. Launch\n\nNovember 21, 2025\n\nUpstage × Karakuri co-developed “Syn Pro” officially certified as a domestic foundation model by Japan’s METI\n\ndocument.addEventListener(\"DOMContentLoaded\", function () {\nconst sets = document.querySelectorAll(\".ab-test-topbar-set\");\n\nif (sets.length === 0) return;\n\nconst randomIndex = Math.floor(Math.random() * sets.length);\n\nsets.forEach((set, idx) => {\nset.style.display = idx === randomIndex ? \"block\" : \"none\"; // 또는 \"inline-flex\", 레이아웃에 따라 조정\n});\n});"},{"ref":"P24","kind":"page","title":"Solar Llm For Writing Project Proposals","date":"2026-06-27T00:01:38.461169+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/solar-llm-for-writing-project-proposals","signal_url":null,"signal_json_url":null,"text":"How to use ‘Solar’ LLM vol. 1 - Tips for writing an effective project proposal \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n‘Solar’ LLM for boosting your work efficiency \nAre you interested in leveraging the capabilities of Large Language Models (LLMs) in your business or various applications? Try ‘Solar’, a small, powerful, and purpose-trained model launched by Upstage! It made a stunning impact on the AI community in December 2023 by achieving first-place on the Open LLM Leaderboard on Hugging Face. Remarkably, the model achieves performance comparable to GPT-3.5 while using fewer parameters and operating at 2.5 times faster speeds. Solar also outshined its competitors in a range of benchmarks, demonstrating its expertise in downsizing LLMs without compromising performance.\n\nUntil March 31st 2024, anyone can experience Solar&#x27;s powerful features through a free API on Upstage Console ! Beyond its impressive efficiency, Solar API Beta offers several compelling features “for free” until its beta promotion ends on March 31st 2024.\nEnglish-Korean translation : With our innovative context-aware approach, this API-accessible version surpasses industry giants like GPT-4 and DeepL in translation quality. Solar&#x27;s ability to understand the broader conversation context, extending beyond individual sentences, leads to superior translation accuracy and nuance.\nChat : Compare Solar’s prowess to other models in the ‘Upstage Console’ . Enjoy seamless conversations and high-quality responses based on your prompt. It has been fine-tuned for multi-turn purposes, showing improved performance in a wide range of natural language processing tasks.\n\nBy now, you might be wondering how to use it for real-world business or work. Let&#x27;s dive into some specific examples.\n1. Writing an outline for a proposal \nAs a project manager, business owner, or professional in the various field, you may need to write a project proposal to present the outlines or plans for your project. It usually includes a description of the project, its o"},{"ref":"P25","kind":"page","title":"Llm Evaluation Part1 Benchmark Datasets","date":"2026-06-27T00:01:38.188041+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/llm-evaluation-part1-benchmark-datasets","signal_url":null,"signal_json_url":null,"text":"LLM evaluation part1. What is a benchmark dataset? \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nIntroduction \nWhy do we need benchmark datasets? \nSince the end of 2022, there has been a surge of new Large Language Models (LLMs) that are accessible to the public. With new LLMs that emerge from here and there, it is getting harder for people to recognize which is the new big deal. So how do we know which one is really good?\nWhat are benchmark datasets? \nBenchmark datasets are like the SAT for LLMs. They are the ​​fixed, standardized approach for assessing the quality of the models. The grades these LLMs get allows us to determine their performance and compare them, and what’s more : know which subjects they are good at. It would be wiser to use a model good at mathematical reasoning on that particular task, instead of any random model that has the best language processing skills.\nAll about benchmark datasets \nThe traditional metrics : Perplexity & BLEU ‍\nEvaluating language models traditionally involves assessing their core capability: predicting the next word in a sequence.\nOne of them is perplexity , measuring the model&#x27;s ability to anticipate upcoming text. A lower perplexity score signifies greater predictive accuracy, reflecting the model&#x27;s proficiency in forecasting the next word. While perplexity is useful for monitoring a model&#x27;s progress during training and to check the basic quality of output, it doesn&#x27;t provide a complete assessment of its appropriateness.\nAnother metric is the BLEU (Bilingual Evaluation Understudy) score. It is used to evaluate how close the output of LM is to the texts created by humans. We can check this by calculating the number of words in the human reference text divided by the total words. BLEU scores range from 0 to 1, with scores closer to 1 indicating greater similarity to human-produced text. However, BLEU has its limitations, as it does not account for the context of the text. For example, a casual text message and a formal news article require different l"},{"ref":"P26","kind":"page","title":"Solar Llm With Predibase The Best Llm For Fine Tuning","date":"2026-06-27T00:01:38.18742+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/solar-llm-with-predibase-the-best-llm-for-fine-tuning","signal_url":null,"signal_json_url":null,"text":"Solar LLM with Predibase: The best LLM for fine-tuning that beats GPT-4 \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nFor organizations with domain-specific data and use cases, fine-tuning is one of the most performant and cost-effective ways to tailor LLMs for production applications. Through fine-tuning small LLMs for specific use cases, teams can achieve performance that surpasses that of massive general models (e.g. GPT-4) for use cases such as:\n\nHowever, not all LLMs are equally efficient for fine-tuning. Some models are more suitable for fine-tuning than others because each model was developed with different design philosophies (e.g., a model designed to be good with broad, general use cases vs. a model designed to be customized for specific applications).\nPredibase is the fastest and most efficient way to fine-tune and serve task-specific LLMs and has deep experience fine-tuning an extensive collection of open-source LLMs and small proprietary LLMs that are ideal for fine-tuning. Upstage and Predibase worked together to discover a faster and more efficient way to fine-tune and serve task-specific LLMs.\nAfter running nearly 500 fine-tuning experiments, we can quantifiably demonstrate that Upstage’s Solar LLM is the most competent model for fine-tuning , and are excited to announce that Solar LLM is now available for teams to fine-tune and serve on Predibase .\nDiscover in-depth reasons why the Solar LLM excels in fine-tuning and explore the platform, designed to enable developers to fine-tune the Solar model.\n[→ Get Started for Free ]\nIntroducing Upstage’s Solar LLMs \nWhy did Upstage build Solar LLM ? \nUpstage is a leading enterprise AI company that has a proven track record of providing powerful custom document processing / LLM solutions for global enterprises across various industries such as financial services, healthcare, supply chain, and legal.\nWith its deep roots in the enterprise AI space, Upstage developed Solar LLM with the belief that for mainstream enterprise adoption, enterprises need a powerful,"},{"ref":"P27","kind":"page","title":"Icdar Win Interview","date":"2026-06-27T00:01:38.171149+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/en/icdar-win-interview","signal_url":null,"signal_json_url":null,"text":"[Starview Vol. 8] ICDAR 2023: AI Startup makes a splash, sweeping four categories \n\nUpstage Studio — Build and deploy your agent.\nLearn more →\nUpstage Studio — Build and deploy your agent.\nLearn more →\n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\nUpstage demonstrated our technical superiority in the field of AI OCR by winning four categories of the prestigious &#x27;ICDAR 2023&#x27; competition, surpassing global tech giants like Amazon, NVIDIA, Alibaba, and Huawei.\nThe &#x27;ICDAR Robust Reading Competition&#x27; evaluates the performance of companies in text detection and recognition from digital images and videos. Upstage&#x27;s exceptional performance resulted in them securing the top position in four categories: HierText-1/2, VQAonBD, and IHTR. Meet our outstanding members behind this remarkable achievement.\nExperience the power of our leading document AI now! → \n\n[HierText-1/2] \nParticipating and focusing on competition is great fun! \nQ: It&#x27;s a pleasure to meet you. Could you introduce yourselves? \nDahyun Kim: Hi, I&#x27;m Dahyun Kim, a multimodal AI researcher at Upstage.\nYoonsoo Kim: Hi, I&#x27;m Yoonsu Kim, a member of the AI Challenges team working on AI research and development.\n\n&#x27;HierText&#x27; extracts everything from textual content in images to their hierarchical structure. Q. Congratulations on your win again! Could you explain the HierText competition? \n\nYoonsoo Kim: HierText is a competition hosted by Google Research that extends beyond detecting words in images and aims to extract hierarchical structures. Extracting hierarchical structures involves detecting words and grouping them into lines, and then grouping lines into paragraphs. In Task 1, participants are evaluated based on their ability to detect hierarchical structures, while in Task 2, they are evaluated on their ability to accurately recognize detected words.\nDaehyun Kim: To clarify further, in Task 1, participants were required to identify the positions of text hierarchically (word, sentence, paragraph) in a given image. In Task 2, participants also had to attempt to read words, essentially performing complete OCR. The main difference lies i"},{"ref":"P28","kind":"page","title":"Artificial Intelligence Examples In Everyday Life","date":"2026-06-27T00:01:38.108172+00:00","date_source":null,"source_url":"https://www.upstage.ai/blog/ko/artificial-intelligence-examples-in-everyday-life","signal_url":null,"signal_json_url":null,"text":"이런 곳에도 인공지능이? 일상 속 인공지능 알아보기 \n\nSolutions\n\nResources\n\nCompany\n\nPricing\n\nTry now\nTry demo\n\nContact us \n\nen \n\n비즈니스의 미래를 선도하는 인공지는 솔루션\n\n비즈니스 고민을 해결할 생성형 AI 모델이 필요하신가요?\n\n업스테이지의 AI전문가와 함께 시작하세요!\n\n✕ \n\n언제부터인가 은행 앱에서 얼굴, 혹은 지문 인식만으로 계좌 이체가 가능해졌습니다. 유행하는 숏클립에서는 자동 생성된 나레이션이 흘러나오죠. 컴퓨터와 몇 가지 대화를 주고 받으면, 눈 깜짝할 새에 아름다운 그림이 탄생하기도 합니다. 나의 대화를 엿듣고 있었던 것만 같은 쇼핑몰의 상품 추천을 보면서, “ 이게 어떻게 가능한 걸까? ”라는 궁금증을 느껴 보신 적이 있나요?\n\n인공지능을 활용한 금융 서비스 (출처: 토스피드 ) ‍\n이 모든 것은 인공지능을 활용한 서비스들입니다. 인공지능은 우리의 디지털 세계에 특별한 힘을 불어 넣고 있습니다. 이 힘으로 우리의 일상이 더 편리해지고, 예측 불가능해 보였던 미래가 조금씩 예측 가능한 형태 로 바뀌어 가고 있습니다. 오늘은 우리 일상을 변화시키고 있는 인공지능 서비스에 대해 이야기해 볼까 합니다. 무거운 기술 용어는 한층 뒤로 미루고, 함께 흥미로운 여정을 떠나볼까요?\n‍\n우리 주변의 다양한 인공지능 서비스 \n네비게이션 🧭 \n\n출처: 네이버지도 공식 블로그 ‍\n초행길의 필수품, 길찾기 서비스! 특히 스마트폰의 여러 애플리케이션을 활용하여 네비게이션을 사용하고 계실텐데요, 이 네비게이션에도 인공지능이 활약하고 있습니다. 출발지와 목적지를 입력하면, 목적지에 도달하는 가장 빠른 경로를 계산해야 하는데, 도로 상황은 시시각각 변하기 때문에 가장 빠른 경로는 매번 바뀔 수 밖에 없습니다. 인공지능은 이렇게 변화하는 도로 상황을 예측 하여 가장 빠른 길을 찾아낼 수 있어요.\n\n인공지능 대화 서비스 💬 \n\n이제는 정말 유명해진 ChatGPT! ChatGPT를 포함하는 대화형 인공지능 서비스는 우리 생활 속에 녹아든 가장 대표적인 인공지능 서비스의 예시입니다. AskUp과 같은 제품을 통해 이미 대중에게 익숙한 플랫폼에서도 대화형 인공지능 서비스를 사용할 수 있죠. “지금 몇시야?”, “오늘 날씨 알려줘.”와 같은 단발성 명령어들을 알아 듣던 대화형 인공지능 서비스는 이제 마치 사람과 대화를 나누는 것과 같은 형태의 챗봇으로 발전 했습니다. 뛰어난 성능과 편리성 덕분에 점점 더 많은 사람들이 손쉽게 대화형 인공지능 서비스를 사용하고 있어요.\n교육 서비스 📖 \n\n(출처: 콴다) ‍\n교육 분야에서도 인공지능이 많이 활용되고 있습니다. 수식을 입력하기만 해도, 알아서 식을 해결하고, 풀이과정과 함께 정답을 알려주는 서비스까지 출시되었죠. 많은 문제 상황에서, 선생님께 찾아가 질문하지 않고도 인공지능이 해설과 정답을 알려줄 수 있습니다. 교사는 단순 문제 풀이에서 벗어나 더욱 학생들에게 집중할 수 있게 되었고, 학생들은 궁금한 것이 있을 때 기다리지 않고 바로 해결할 수 있게 되었죠. 업스테이지에서는 LLM을 활용해서 환각 없이 정확한 해설로 수학 문제를 풀어내는 수학 특화 Private LLM을 콴다와 만들어가고 있습니다. >> 자세히 보러가기 \n‍\n검색 서비스 🔍 \n\n사용자가 여러 페이지를 클릭하며 이동하지 않고 원하는 결과를 바로 받을 수 있는 Zero-click search ‍\n포털사이트에서 검색을 하면 연관 검색어가 수십 페이지에 걸쳐 나오던 과거의 검색 서비스는 인공지능이 도입됨에 따라 그 패러다임이 크게 변화하였습니다. 검색어를 입력하면, 사용자가 여러 페이지를 클릭하며 이동하지 않고 바로 원하는 결과를 받을 수 있는 zero-click search 가 활성화되었기 때문이에요. 검색창에 단어를 입력하고, 원하는 사이트로 이동하기 위해 페이지들을 헤매던 시절은 이제 추억 속으로 사라지고 있어요.\n‍\n자율주행 서비스 🚗 \n\n출처: 테슬라 ‍\n자율주행 서비스는 운전자의 개입 없이 차량이 스스로 주행을 수행 하는 혁신적인 기술입니다. 다양한 센서와 카메라, 레이더 등을 통해 주변 환경을 실시간으로 파악하고 분석하며, 분석한 데이터를 인공지능 알고리즘이 처리하여 도로 상황을 예측하고 적절한 주행 전략을 수립합니다. 언젠가 모든 차량이 자율주행 기능을 탑재하면, 교통 흐름이 최적화되어 교통사고 예방뿐 아니라 운전에 제약이 있는 사람들에게 독립적인 이동 수단을 제공하여 삶의 질을 전반적으로 향상시"},{"ref":"E1","kind":"event","title":"upstage/SOLAR-10.7B-Instruct-v1.0","date":"2023-12-12T12:39:22+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0","signal_url":"https://onlylabs.fyi/signals/d66e12c7-6583-4627-9f46-fe5462b3f6b6","signal_json_url":"https://onlylabs.fyi/signals/d66e12c7-6583-4627-9f46-fe5462b3f6b6/signal.json","text":"model_released · upstage/SOLAR-10.7B-Instruct-v1.0 · signal_desk=releases · occurred_at=2023-12-12T12:39:22+00:00 · url=https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0 · hf_downloads=63564 · hf_likes=656 · hf_params=10731524096 · pipeline=text-generation · license=cc-by-nc-4.0 · raw={\"derived_reason\":\"first-party-finetune\"}"},{"ref":"E2","kind":"event","title":"upstage/Solar-Open-100B","date":"2025-12-10T01:10:35+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/Solar-Open-100B","signal_url":"https://onlylabs.fyi/signals/5f3e4860-16d6-403c-8200-7cdd2ec64fa8","signal_json_url":"https://onlylabs.fyi/signals/5f3e4860-16d6-403c-8200-7cdd2ec64fa8/signal.json","text":"model_released · upstage/Solar-Open-100B · signal_desk=releases · occurred_at=2025-12-10T01:10:35+00:00 · url=https://huggingface.co/upstage/Solar-Open-100B · hf_downloads=5056 · hf_likes=479 · hf_params=102651793408 · pipeline=text-generation · license=other"},{"ref":"E3","kind":"event","title":"upstage/solar-pro-preview-instruct","date":"2024-09-09T01:08:58+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro-preview-instruct","signal_url":"https://onlylabs.fyi/signals/316469cd-f4d4-49c6-8b0d-70b369c67918","signal_json_url":"https://onlylabs.fyi/signals/316469cd-f4d4-49c6-8b0d-70b369c67918/signal.json","text":"model_released · upstage/solar-pro-preview-instruct · signal_desk=releases · occurred_at=2024-09-09T01:08:58+00:00 · url=https://huggingface.co/upstage/solar-pro-preview-instruct · hf_downloads=42736 · hf_likes=457 · hf_params=22140032000 · pipeline=text-generation · license=mit"},{"ref":"E4","kind":"event","title":"upstage/SOLAR-10.7B-v1.0","date":"2023-12-12T14:57:41+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/SOLAR-10.7B-v1.0","signal_url":"https://onlylabs.fyi/signals/1db38aca-a53f-4179-b043-2e2ecef28731","signal_json_url":"https://onlylabs.fyi/signals/1db38aca-a53f-4179-b043-2e2ecef28731/signal.json","text":"model_released · upstage/SOLAR-10.7B-v1.0 · signal_desk=releases · occurred_at=2023-12-12T14:57:41+00:00 · url=https://huggingface.co/upstage/SOLAR-10.7B-v1.0 · hf_downloads=12096 · hf_likes=321 · hf_params=10731524096 · pipeline=text-generation · license=apache-2.0"},{"ref":"E5","kind":"event","title":"upstage/SOLAR-0-70b-16bit","date":"2023-07-30T01:10:53+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/SOLAR-0-70b-16bit","signal_url":"https://onlylabs.fyi/signals/30dbdd3d-4b2e-49d2-b83b-fe818eebe5b4","signal_json_url":"https://onlylabs.fyi/signals/30dbdd3d-4b2e-49d2-b83b-fe818eebe5b4/signal.json","text":"model_released · upstage/SOLAR-0-70b-16bit · signal_desk=releases · occurred_at=2023-07-30T01:10:53+00:00 · url=https://huggingface.co/upstage/SOLAR-0-70b-16bit · hf_downloads=13132 · hf_likes=259 · pipeline=text-generation"},{"ref":"E6","kind":"event","title":"upstage/llama-30b-instruct-2048","date":"2023-07-13T12:06:18+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/llama-30b-instruct-2048","signal_url":"https://onlylabs.fyi/signals/5f7e7cad-ebcc-4d87-b176-8b4696122a88","signal_json_url":"https://onlylabs.fyi/signals/5f7e7cad-ebcc-4d87-b176-8b4696122a88/signal.json","text":"model_released · upstage/llama-30b-instruct-2048 · signal_desk=releases · occurred_at=2023-07-13T12:06:18+00:00 · url=https://huggingface.co/upstage/llama-30b-instruct-2048 · hf_downloads=892 · hf_likes=103 · pipeline=text-generation"},{"ref":"E7","kind":"event","title":"upstage/Llama-2-70b-instruct","date":"2023-07-24T09:13:08+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/Llama-2-70b-instruct","signal_url":"https://onlylabs.fyi/signals/8d1e5d50-95ca-4512-b0a3-a794275f787e","signal_json_url":"https://onlylabs.fyi/signals/8d1e5d50-95ca-4512-b0a3-a794275f787e/signal.json","text":"model_released · upstage/Llama-2-70b-instruct · signal_desk=releases · occurred_at=2023-07-24T09:13:08+00:00 · url=https://huggingface.co/upstage/Llama-2-70b-instruct · hf_downloads=1006 · hf_likes=63 · pipeline=text-generation"},{"ref":"E8","kind":"event","title":"upstage/solar-pro-preview-pretrained","date":"2024-08-29T21:29:29+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro-preview-pretrained","signal_url":"https://onlylabs.fyi/signals/abd57a06-ece0-4789-8648-a5c027fc1bd3","signal_json_url":"https://onlylabs.fyi/signals/abd57a06-ece0-4789-8648-a5c027fc1bd3/signal.json","text":"model_released · upstage/solar-pro-preview-pretrained · signal_desk=releases · occurred_at=2024-08-29T21:29:29+00:00 · url=https://huggingface.co/upstage/solar-pro-preview-pretrained · hf_likes=62 · hf_params=22140032000 · pipeline=text-generation"},{"ref":"E9","kind":"event","title":"Ai Powered Proofreading For Professionals","date":"2026-06-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/ko/ai-powered-proofreading-for-professionals","signal_url":"https://onlylabs.fyi/signals/a864dc2a-fd94-4d92-ad65-1427ce19d96d","signal_json_url":"https://onlylabs.fyi/signals/a864dc2a-fd94-4d92-ad65-1427ce19d96d/signal.json","text":"post_published · Ai Powered Proofreading For Professionals · signal_desk=talking · occurred_at=2026-06-26T00:00:00.000Z · url=https://www.upstage.ai/blog/ko/ai-powered-proofreading-for-professionals"},{"ref":"E10","kind":"event","title":"2pyeon Api Ereokodeu Wanjeon Haebu","date":"2026-06-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/ko/2pyeon-api-ereokodeu-wanjeon-haebu","signal_url":"https://onlylabs.fyi/signals/edaa9519-0d66-47d2-bfee-cfe44b52ef01","signal_json_url":"https://onlylabs.fyi/signals/edaa9519-0d66-47d2-bfee-cfe44b52ef01/signal.json","text":"post_published · 2pyeon Api Ereokodeu Wanjeon Haebu · signal_desk=talking · occurred_at=2026-06-26T00:00:00.000Z · url=https://www.upstage.ai/blog/ko/2pyeon-api-ereokodeu-wanjeon-haebu"},{"ref":"E11","kind":"event","title":"Ai Space Real Work Automation Guide","date":"2026-06-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/ko/ai-space-real-work-automation-guide","signal_url":"https://onlylabs.fyi/signals/a8f14d64-0a4c-4cbd-b314-a40b3eee8606","signal_json_url":"https://onlylabs.fyi/signals/a8f14d64-0a4c-4cbd-b314-a40b3eee8606/signal.json","text":"post_published · Ai Space Real Work Automation Guide · signal_desk=talking · occurred_at=2026-06-26T00:00:00.000Z · url=https://www.upstage.ai/blog/ko/ai-space-real-work-automation-guide"},{"ref":"E12","kind":"event","title":"Askup Use Case Health","date":"2026-06-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/ko/askup-use-case-health","signal_url":"https://onlylabs.fyi/signals/d8248a52-fab6-4d86-a471-6dca74194494","signal_json_url":"https://onlylabs.fyi/signals/d8248a52-fab6-4d86-a471-6dca74194494/signal.json","text":"post_published · Askup Use Case Health · signal_desk=talking · occurred_at=2026-06-26T00:00:00.000Z · url=https://www.upstage.ai/blog/ko/askup-use-case-health"},{"ref":"E13","kind":"event","title":"Askup Use Case Travel","date":"2026-06-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/ko/askup-use-case-travel","signal_url":"https://onlylabs.fyi/signals/1d34643d-de46-4699-82af-8fabcdd93c69","signal_json_url":"https://onlylabs.fyi/signals/1d34643d-de46-4699-82af-8fabcdd93c69/signal.json","text":"post_published · Askup Use Case Travel · signal_desk=talking · occurred_at=2026-06-26T00:00:00.000Z · url=https://www.upstage.ai/blog/ko/askup-use-case-travel"},{"ref":"E14","kind":"event","title":"UpstageAI/knu-2026-summer-bootcamp-project","date":"2026-06-12T07:06:29+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/knu-2026-summer-bootcamp-project","signal_url":"https://onlylabs.fyi/signals/de6cbfa6-4b9a-4bdb-8054-19198539fe12","signal_json_url":"https://onlylabs.fyi/signals/de6cbfa6-4b9a-4bdb-8054-19198539fe12/signal.json","text":"repo_new · UpstageAI/knu-2026-summer-bootcamp-project · signal_desk=repos · occurred_at=2026-06-12T07:06:29+00:00 · url=https://github.com/UpstageAI/knu-2026-summer-bootcamp-project · raw={\"repo\":\"UpstageAI/knu-2026-summer-bootcamp-project\",\"description\":\"강원대학교 2026 여름학기 부트캠프 최종 프로젝트 제출 (마감: 2026-07-14 12:00)\",\"language\":\"Python\"}"},{"ref":"E15","kind":"event","title":"UpstageAI/hermes-agent","date":"2026-06-08T02:36:18+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/hermes-agent","signal_url":"https://onlylabs.fyi/signals/559043f9-080e-4c7f-9629-6db81b6f7215","signal_json_url":"https://onlylabs.fyi/signals/559043f9-080e-4c7f-9629-6db81b6f7215/signal.json","text":"repo_forked · UpstageAI/hermes-agent · signal_desk=forks · occurred_at=2026-06-08T02:36:18+00:00 · url=https://github.com/UpstageAI/hermes-agent · raw={\"repo\":\"UpstageAI/hermes-agent\",\"parent\":\"NousResearch/hermes-agent\"}"},{"ref":"E16","kind":"event","title":"upstage/TinySolar-187m-4k","date":"2026-05-27T03:11:31+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-187m-4k","signal_url":"https://onlylabs.fyi/signals/d2a23d04-ba4a-4bc1-9390-4a43e15bcfaf","signal_json_url":"https://onlylabs.fyi/signals/d2a23d04-ba4a-4bc1-9390-4a43e15bcfaf/signal.json","text":"model_released · upstage/TinySolar-187m-4k · signal_desk=releases · occurred_at=2026-05-27T03:11:31+00:00 · url=https://huggingface.co/upstage/TinySolar-187m-4k · hf_downloads=76 · hf_likes=3 · hf_params=187188224 · license=apache-2.0"},{"ref":"E17","kind":"event","title":"upstage/TinySolar-111m-4k","date":"2026-05-27T03:11:21+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-111m-4k","signal_url":"https://onlylabs.fyi/signals/5b43238d-0447-4885-a3dc-64c5b43beb64","signal_json_url":"https://onlylabs.fyi/signals/5b43238d-0447-4885-a3dc-64c5b43beb64/signal.json","text":"model_released · upstage/TinySolar-111m-4k · signal_desk=releases · occurred_at=2026-05-27T03:11:21+00:00 · url=https://huggingface.co/upstage/TinySolar-111m-4k · hf_downloads=20 · hf_likes=3 · hf_params=111156224 · license=apache-2.0"},{"ref":"E18","kind":"event","title":"UpstageAI/upstage-ambassador-2nd-final-project","date":"2026-05-27T04:05:19+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/upstage-ambassador-2nd-final-project","signal_url":"https://onlylabs.fyi/signals/edc79978-c686-4eb1-9ed4-90838928c967","signal_json_url":"https://onlylabs.fyi/signals/edc79978-c686-4eb1-9ed4-90838928c967/signal.json","text":"repo_new · UpstageAI/upstage-ambassador-2nd-final-project · signal_desk=repos · occurred_at=2026-05-27T04:05:19+00:00 · url=https://github.com/UpstageAI/upstage-ambassador-2nd-final-project · raw={\"repo\":\"UpstageAI/upstage-ambassador-2nd-final-project\",\"description\":\"업스테이지 앰버서더 2기 최종 프로젝트 제출 저장소\",\"language\":\"Python\"}"},{"ref":"E19","kind":"event","title":"UpstageAI/seoul-sw-maestro-project","date":"2026-05-21T01:10:39+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/seoul-sw-maestro-project","signal_url":"https://onlylabs.fyi/signals/b8e15867-82c2-490e-ad65-8d89cd761384","signal_json_url":"https://onlylabs.fyi/signals/b8e15867-82c2-490e-ad65-8d89cd761384/signal.json","text":"repo_new · UpstageAI/seoul-sw-maestro-project · signal_desk=repos · occurred_at=2026-05-21T01:10:39+00:00 · url=https://github.com/UpstageAI/seoul-sw-maestro-project · raw={\"repo\":\"UpstageAI/seoul-sw-maestro-project\",\"description\":\"SW Maestro project submission scaffold\"}"},{"ref":"E20","kind":"event","title":"UpstageAI/busan-sw-maestro-project","date":"2026-05-21T01:10:34+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/busan-sw-maestro-project","signal_url":"https://onlylabs.fyi/signals/0dc521e0-898a-4466-ba69-f59fb98616f7","signal_json_url":"https://onlylabs.fyi/signals/0dc521e0-898a-4466-ba69-f59fb98616f7/signal.json","text":"repo_new · UpstageAI/busan-sw-maestro-project · signal_desk=repos · occurred_at=2026-05-21T01:10:34+00:00 · url=https://github.com/UpstageAI/busan-sw-maestro-project · raw={\"repo\":\"UpstageAI/busan-sw-maestro-project\",\"description\":\"SW Maestro project submission scaffold\",\"language\":\"Python\"}"},{"ref":"E21","kind":"event","title":"upstage/llama-30b-instruct","date":"2023-07-11T02:41:53+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/llama-30b-instruct","signal_url":"https://onlylabs.fyi/signals/136feb42-bed4-4ae5-9a67-3a5f91a4298e","signal_json_url":"https://onlylabs.fyi/signals/136feb42-bed4-4ae5-9a67-3a5f91a4298e/signal.json","text":"model_released · upstage/llama-30b-instruct · signal_desk=releases · occurred_at=2023-07-11T02:41:53+00:00 · url=https://huggingface.co/upstage/llama-30b-instruct · hf_downloads=900 · hf_likes=24 · pipeline=text-generation"},{"ref":"E22","kind":"event","title":"UpstageAI/edu-usecase-hackathon","date":"2026-05-16T04:23:18+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/edu-usecase-hackathon","signal_url":"https://onlylabs.fyi/signals/a0b1ee78-a24b-402f-a9d7-4707bac72871","signal_json_url":"https://onlylabs.fyi/signals/a0b1ee78-a24b-402f-a9d7-4707bac72871/signal.json","text":"repo_new · UpstageAI/edu-usecase-hackathon · signal_desk=repos · occurred_at=2026-05-16T04:23:18+00:00 · url=https://github.com/UpstageAI/edu-usecase-hackathon · raw={\"repo\":\"UpstageAI/edu-usecase-hackathon\",\"description\":\"해커톤 유즈케이스 \"}"},{"ref":"E23","kind":"event","title":"upstage/llama-65b-instruct","date":"2023-07-17T12:24:11+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/llama-65b-instruct","signal_url":"https://onlylabs.fyi/signals/f70add8e-3e66-4a7c-b597-a8a7e5412626","signal_json_url":"https://onlylabs.fyi/signals/f70add8e-3e66-4a7c-b597-a8a7e5412626/signal.json","text":"model_released · upstage/llama-65b-instruct · signal_desk=releases · occurred_at=2023-07-17T12:24:11+00:00 · url=https://huggingface.co/upstage/llama-65b-instruct · hf_downloads=805 · hf_likes=15 · pipeline=text-generation"},{"ref":"E24","kind":"event","title":"upstage/solar-1-mini-tokenizer","date":"2024-05-02T06:26:22+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-1-mini-tokenizer","signal_url":"https://onlylabs.fyi/signals/f3f2f097-6a68-4a37-9d4e-1731f4ed959f","signal_json_url":"https://onlylabs.fyi/signals/f3f2f097-6a68-4a37-9d4e-1731f4ed959f/signal.json","text":"model_released · upstage/solar-1-mini-tokenizer · signal_desk=releases · occurred_at=2024-05-02T06:26:22+00:00 · url=https://huggingface.co/upstage/solar-1-mini-tokenizer · hf_likes=14 · license=apache-2.0"},{"ref":"E25","kind":"event","title":"upstage/TinySolar-248m-4k","date":"2024-02-07T14:49:15+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-248m-4k","signal_url":"https://onlylabs.fyi/signals/e97e6ef2-0a31-4a16-a011-d5cb795b3415","signal_json_url":"https://onlylabs.fyi/signals/e97e6ef2-0a31-4a16-a011-d5cb795b3415/signal.json","text":"model_released · upstage/TinySolar-248m-4k · signal_desk=releases · occurred_at=2024-02-07T14:49:15+00:00 · url=https://huggingface.co/upstage/TinySolar-248m-4k · hf_downloads=417 · hf_likes=12 · hf_params=248013824 · pipeline=text-generation · license=apache-2.0"},{"ref":"E26","kind":"event","title":"upstage/solar-pro-preview-tokenizer","date":"2024-09-04T01:59:07+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro-preview-tokenizer","signal_url":"https://onlylabs.fyi/signals/ba21775f-e133-47f1-bd4e-6cd41b7eec06","signal_json_url":"https://onlylabs.fyi/signals/ba21775f-e133-47f1-bd4e-6cd41b7eec06/signal.json","text":"model_released · upstage/solar-pro-preview-tokenizer · signal_desk=releases · occurred_at=2024-09-04T01:59:07+00:00 · url=https://huggingface.co/upstage/solar-pro-preview-tokenizer · hf_likes=12 · license=apache-2.0"},{"ref":"E27","kind":"event","title":"UpstageAI/upstage-extensions-hub","date":"2026-04-20T04:21:54+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/upstage-extensions-hub","signal_url":"https://onlylabs.fyi/signals/becf30e8-6640-4619-bf99-abfee14e2952","signal_json_url":"https://onlylabs.fyi/signals/becf30e8-6640-4619-bf99-abfee14e2952/signal.json","text":"repo_new · UpstageAI/upstage-extensions-hub · signal_desk=repos · occurred_at=2026-04-20T04:21:54+00:00 · url=https://github.com/UpstageAI/upstage-extensions-hub · stars=9 · raw={\"repo\":\"UpstageAI/upstage-extensions-hub\",\"language\":\"Shell\"}"},{"ref":"E28","kind":"event","title":"upstage/TinySolar-248m-4k-code-instruct","date":"2024-04-15T02:21:14+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-248m-4k-code-instruct","signal_url":"https://onlylabs.fyi/signals/26fd98a4-8ec3-464b-a92b-edf69639afe3","signal_json_url":"https://onlylabs.fyi/signals/26fd98a4-8ec3-464b-a92b-edf69639afe3/signal.json","text":"model_released · upstage/TinySolar-248m-4k-code-instruct · signal_desk=releases · occurred_at=2024-04-15T02:21:14+00:00 · url=https://huggingface.co/upstage/TinySolar-248m-4k-code-instruct · hf_downloads=107 · hf_likes=9 · hf_params=248013824 · pipeline=text-generation · license=apache-2.0"},{"ref":"E29","kind":"event","title":"upstage/solar-docvision-preview-tokenizer","date":"2024-09-04T07:15:35+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-docvision-preview-tokenizer","signal_url":"https://onlylabs.fyi/signals/5301dbb0-9552-4f38-a58f-6bd2d2e92f23","signal_json_url":"https://onlylabs.fyi/signals/5301dbb0-9552-4f38-a58f-6bd2d2e92f23/signal.json","text":"model_released · upstage/solar-docvision-preview-tokenizer · signal_desk=releases · occurred_at=2024-09-04T07:15:35+00:00 · url=https://huggingface.co/upstage/solar-docvision-preview-tokenizer · hf_likes=7 · license=apache-2.0"},{"ref":"E30","kind":"event","title":"upstage/TinySolar-248m-4k-py","date":"2024-02-07T03:33:41+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-248m-4k-py","signal_url":"https://onlylabs.fyi/signals/d4cb7cc4-ead9-457e-8f2d-43839c05f158","signal_json_url":"https://onlylabs.fyi/signals/d4cb7cc4-ead9-457e-8f2d-43839c05f158/signal.json","text":"model_released · upstage/TinySolar-248m-4k-py · signal_desk=releases · occurred_at=2024-02-07T03:33:41+00:00 · url=https://huggingface.co/upstage/TinySolar-248m-4k-py · hf_downloads=106 · hf_likes=5 · hf_params=248013824 · pipeline=text-generation · license=apache-2.0"},{"ref":"E31","kind":"event","title":"upstage/solar-pro-tokenizer","date":"2024-11-15T05:14:07+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro-tokenizer","signal_url":"https://onlylabs.fyi/signals/e279a0c1-088d-4e02-8eff-513937482b5d","signal_json_url":"https://onlylabs.fyi/signals/e279a0c1-088d-4e02-8eff-513937482b5d/signal.json","text":"model_released · upstage/solar-pro-tokenizer · signal_desk=releases · occurred_at=2024-11-15T05:14:07+00:00 · url=https://huggingface.co/upstage/solar-pro-tokenizer · hf_likes=4 · license=apache-2.0"},{"ref":"E32","kind":"event","title":"UpstageAI/dataverse","date":"2023-08-21T15:50:11+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/dataverse","signal_url":"https://onlylabs.fyi/signals/836c47a0-11aa-4634-8493-a4f24384087b","signal_json_url":"https://onlylabs.fyi/signals/836c47a0-11aa-4634-8493-a4f24384087b/signal.json","text":"repo_new · UpstageAI/dataverse · signal_desk=repos · occurred_at=2023-08-21T15:50:11+00:00 · url=https://github.com/UpstageAI/dataverse · stars=563 · raw={\"repo\":\"UpstageAI/dataverse\",\"description\":\"The Universe of Data. All about data, data science, and data engineering\",\"language\":\"Python\"}"},{"ref":"E33","kind":"event","title":"upstage/TinySolar-248m-4k-py-instruct","date":"2024-04-11T08:33:20+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TinySolar-248m-4k-py-instruct","signal_url":"https://onlylabs.fyi/signals/d24f783d-ff2c-491e-9904-7f11d6cbdb26","signal_json_url":"https://onlylabs.fyi/signals/d24f783d-ff2c-491e-9904-7f11d6cbdb26/signal.json","text":"model_released · upstage/TinySolar-248m-4k-py-instruct · signal_desk=releases · occurred_at=2024-04-11T08:33:20+00:00 · url=https://huggingface.co/upstage/TinySolar-248m-4k-py-instruct · hf_downloads=68 · hf_likes=2 · hf_params=248013824 · pipeline=text-generation · license=apache-2.0"},{"ref":"E34","kind":"event","title":"upstage/TFLOP","date":"2025-11-06T06:45:38+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/TFLOP","signal_url":"https://onlylabs.fyi/signals/f3a6e949-4181-448d-9d4e-90d73ddc5fd4","signal_json_url":"https://onlylabs.fyi/signals/f3a6e949-4181-448d-9d4e-90d73ddc5fd4/signal.json","text":"model_released · upstage/TFLOP · signal_desk=releases · occurred_at=2025-11-06T06:45:38+00:00 · url=https://huggingface.co/upstage/TFLOP · hf_downloads=24 · hf_likes=2 · license=cc-by-nc-4.0"},{"ref":"E35","kind":"event","title":"upstage/solar-pro3-tokenizer","date":"2026-01-23T08:41:27+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro3-tokenizer","signal_url":"https://onlylabs.fyi/signals/39369cc6-2acb-4c22-a023-8aacde9721df","signal_json_url":"https://onlylabs.fyi/signals/39369cc6-2acb-4c22-a023-8aacde9721df/signal.json","text":"model_released · upstage/solar-pro3-tokenizer · signal_desk=releases · occurred_at=2026-01-23T08:41:27+00:00 · url=https://huggingface.co/upstage/solar-pro3-tokenizer · hf_likes=2 · license=other"},{"ref":"E36","kind":"event","title":"UpstageAI/solar-prompt-cookbook","date":"2024-11-04T07:11:37+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/solar-prompt-cookbook","signal_url":"https://onlylabs.fyi/signals/df0c2aa9-f2d3-4491-a944-a4f17f551e3e","signal_json_url":"https://onlylabs.fyi/signals/df0c2aa9-f2d3-4491-a944-a4f17f551e3e/signal.json","text":"repo_new · UpstageAI/solar-prompt-cookbook · signal_desk=repos · occurred_at=2024-11-04T07:11:37+00:00 · url=https://github.com/UpstageAI/solar-prompt-cookbook · stars=252 · raw={\"repo\":\"UpstageAI/solar-prompt-cookbook\",\"description\":\"Solar Prompt Cookbook\",\"language\":\"Jupyter Notebook\"}"},{"ref":"E37","kind":"event","title":"UpstageAI/evalverse","date":"2024-03-22T00:11:47+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/evalverse","signal_url":"https://onlylabs.fyi/signals/1c5c1920-1c1d-42bb-b74f-ef7881e10a42","signal_json_url":"https://onlylabs.fyi/signals/1c5c1920-1c1d-42bb-b74f-ef7881e10a42/signal.json","text":"repo_new · UpstageAI/evalverse · signal_desk=repos · occurred_at=2024-03-22T00:11:47+00:00 · url=https://github.com/UpstageAI/evalverse · stars=235 · raw={\"repo\":\"UpstageAI/evalverse\",\"description\":\"The Universe of Evaluation. All about the evaluation for LLMs.\",\"language\":\"Python\"}"},{"ref":"E38","kind":"event","title":"UpstageAI/security-audit","date":"2026-04-01T03:36:08+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/security-audit","signal_url":"https://onlylabs.fyi/signals/c05b141b-94ce-43d5-bf49-dd6edf776053","signal_json_url":"https://onlylabs.fyi/signals/c05b141b-94ce-43d5-bf49-dd6edf776053/signal.json","text":"repo_new · UpstageAI/security-audit · signal_desk=repos · occurred_at=2026-04-01T03:36:08+00:00 · url=https://github.com/UpstageAI/security-audit · raw={\"repo\":\"UpstageAI/security-audit\",\"description\":\"Public\",\"language\":\"Shell\"}"},{"ref":"E39","kind":"event","title":"upstage/solar-pro2-tokenizer","date":"2025-07-08T03:31:42+00:00","date_source":"source","source_url":"https://huggingface.co/upstage/solar-pro2-tokenizer","signal_url":"https://onlylabs.fyi/signals/06fc6466-0696-4f42-a8ce-1c0f491e979e","signal_json_url":"https://onlylabs.fyi/signals/06fc6466-0696-4f42-a8ce-1c0f491e979e/signal.json","text":"model_released · upstage/solar-pro2-tokenizer · signal_desk=releases · occurred_at=2025-07-08T03:31:42+00:00 · url=https://huggingface.co/upstage/solar-pro2-tokenizer · hf_likes=1 · license=other"},{"ref":"E40","kind":"event","title":"UpstageAI/cookbook","date":"2024-05-21T05:45:00+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/cookbook","signal_url":"https://onlylabs.fyi/signals/fff2fefc-6b1f-47fd-9a51-0cea477f89e2","signal_json_url":"https://onlylabs.fyi/signals/fff2fefc-6b1f-47fd-9a51-0cea477f89e2/signal.json","text":"repo_new · UpstageAI/cookbook · signal_desk=repos · occurred_at=2024-05-21T05:45:00+00:00 · url=https://github.com/UpstageAI/cookbook · stars=199 · raw={\"repo\":\"UpstageAI/cookbook\",\"description\":\"Upstage api examples and guides\",\"language\":\"Jupyter Notebook\"}"},{"ref":"E41","kind":"event","title":"UpstageAI/TFLOP","date":"2024-04-22T10:48:10+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/TFLOP","signal_url":"https://onlylabs.fyi/signals/4007f7eb-97e5-4e4f-967a-19bfb7eaa61a","signal_json_url":"https://onlylabs.fyi/signals/4007f7eb-97e5-4e4f-967a-19bfb7eaa61a/signal.json","text":"repo_new · UpstageAI/TFLOP · signal_desk=repos · occurred_at=2024-04-22T10:48:10+00:00 · url=https://github.com/UpstageAI/TFLOP · stars=51 · raw={\"repo\":\"UpstageAI/TFLOP\",\"description\":\"Official Implementation of TFLOP: Table Structure Recognition Framework with Layout Pointer Mechanism\",\"language\":\"Python\"}"},{"ref":"E42","kind":"event","title":"UpstageAI/2022-lguplus-AI-Ground","date":"2022-10-06T02:10:59+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/2022-lguplus-AI-Ground","signal_url":"https://onlylabs.fyi/signals/5ccfefa8-ceb9-475c-a91b-48602e23e1dc","signal_json_url":"https://onlylabs.fyi/signals/5ccfefa8-ceb9-475c-a91b-48602e23e1dc/signal.json","text":"repo_new · UpstageAI/2022-lguplus-AI-Ground · signal_desk=repos · occurred_at=2022-10-06T02:10:59+00:00 · url=https://github.com/UpstageAI/2022-lguplus-AI-Ground · stars=49 · raw={\"repo\":\"UpstageAI/2022-lguplus-AI-Ground\"}"},{"ref":"E43","kind":"event","title":"UpstageAI/cl4kt","date":"2022-02-02T12:09:03+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/cl4kt","signal_url":"https://onlylabs.fyi/signals/7856e4cc-725b-4e25-be06-394210d5a7ef","signal_json_url":"https://onlylabs.fyi/signals/7856e4cc-725b-4e25-be06-394210d5a7ef/signal.json","text":"repo_new · UpstageAI/cl4kt · signal_desk=repos · occurred_at=2022-02-02T12:09:03+00:00 · url=https://github.com/UpstageAI/cl4kt · stars=31 · raw={\"repo\":\"UpstageAI/cl4kt\",\"description\":\"This is our implementation for the paper \\\"Contrastive Learning for Knowledge Tracing\\\" (TheWebConf 2022).\",\"language\":\"Python\"}"},{"ref":"E44","kind":"event","title":"UpstageAI/upstage-open-chat","date":"2025-04-08T22:09:24+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/upstage-open-chat","signal_url":"https://onlylabs.fyi/signals/9aaf9d88-7460-4687-95cf-8519e5cbb887","signal_json_url":"https://onlylabs.fyi/signals/9aaf9d88-7460-4687-95cf-8519e5cbb887/signal.json","text":"repo_forked · UpstageAI/upstage-open-chat · signal_desk=forks · occurred_at=2025-04-08T22:09:24+00:00 · url=https://github.com/UpstageAI/upstage-open-chat · stars=4 · raw={\"repo\":\"UpstageAI/upstage-open-chat\",\"parent\":\"open-webui/open-webui\"}"},{"ref":"E45","kind":"event","title":"UpstageAI/vllm","date":"2025-12-29T06:50:24+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/vllm","signal_url":"https://onlylabs.fyi/signals/176d8916-1297-4814-bdbc-81b010962da6","signal_json_url":"https://onlylabs.fyi/signals/176d8916-1297-4814-bdbc-81b010962da6/signal.json","text":"repo_forked · UpstageAI/vllm · signal_desk=forks · occurred_at=2025-12-29T06:50:24+00:00 · url=https://github.com/UpstageAI/vllm · stars=1 · raw={\"repo\":\"UpstageAI/vllm\",\"parent\":\"vllm-project/vllm\"}"},{"ref":"E46","kind":"event","title":"UpstageAI/gpushare-scheduler-extender","date":"2025-01-31T01:21:54+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/gpushare-scheduler-extender","signal_url":"https://onlylabs.fyi/signals/3da798ae-8ce7-481b-9b2d-68a3b9ba82ea","signal_json_url":"https://onlylabs.fyi/signals/3da798ae-8ce7-481b-9b2d-68a3b9ba82ea/signal.json","text":"repo_forked · UpstageAI/gpushare-scheduler-extender · signal_desk=forks · occurred_at=2025-01-31T01:21:54+00:00 · url=https://github.com/UpstageAI/gpushare-scheduler-extender · stars=1 · raw={\"repo\":\"UpstageAI/gpushare-scheduler-extender\",\"parent\":\"AliyunContainerService/gpushare-scheduler-extender\"}"},{"ref":"E47","kind":"event","title":"Solar Pro 2","date":"2026-03-19T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/news/solar-pro-2","signal_url":"https://onlylabs.fyi/signals/700c64fb-0b01-4153-8dfe-a64bd079e0ec","signal_json_url":"https://onlylabs.fyi/signals/700c64fb-0b01-4153-8dfe-a64bd079e0ec/signal.json","text":"post_published · Solar Pro 2 · signal_desk=talking · occurred_at=2026-03-19T00:00:00.000Z · url=https://www.upstage.ai/news/solar-pro-2"},{"ref":"E48","kind":"event","title":"Upstage Furiosa Llm","date":"2026-03-19T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/news/upstage-furiosa-llm","signal_url":"https://onlylabs.fyi/signals/f9548040-099f-4475-8dbf-cec09a33f8d9","signal_json_url":"https://onlylabs.fyi/signals/f9548040-099f-4475-8dbf-cec09a33f8d9/signal.json","text":"post_published · Upstage Furiosa Llm · signal_desk=talking · occurred_at=2026-03-19T00:00:00.000Z · url=https://www.upstage.ai/news/upstage-furiosa-llm"},{"ref":"E49","kind":"event","title":"Upstage Amd Partnership","date":"2026-03-18T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/news/upstage-amd-partnership","signal_url":"https://onlylabs.fyi/signals/8d035ae9-a9b0-4108-862f-1ee4d033c4a0","signal_json_url":"https://onlylabs.fyi/signals/8d035ae9-a9b0-4108-862f-1ee4d033c4a0/signal.json","text":"post_published · Upstage Amd Partnership · signal_desk=talking · occurred_at=2026-03-18T00:00:00.000Z · url=https://www.upstage.ai/news/upstage-amd-partnership"},{"ref":"E50","kind":"event","title":"UpstageAI/opencode","date":"2026-02-22T20:57:19+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/opencode","signal_url":"https://onlylabs.fyi/signals/df5240f5-a7c7-4253-85e9-512bcba041d9","signal_json_url":"https://onlylabs.fyi/signals/df5240f5-a7c7-4253-85e9-512bcba041d9/signal.json","text":"repo_forked · UpstageAI/opencode · signal_desk=forks · occurred_at=2026-02-22T20:57:19+00:00 · url=https://github.com/UpstageAI/opencode · raw={\"repo\":\"UpstageAI/opencode\",\"parent\":\"anomalyco/opencode\"}"},{"ref":"E51","kind":"event","title":"UpstageAI/Gym","date":"2026-01-31T02:52:48+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/Gym","signal_url":"https://onlylabs.fyi/signals/96714acf-bfc6-4f25-ad4d-dcb3e019eaae","signal_json_url":"https://onlylabs.fyi/signals/96714acf-bfc6-4f25-ad4d-dcb3e019eaae/signal.json","text":"repo_forked · UpstageAI/Gym · signal_desk=forks · occurred_at=2026-01-31T02:52:48+00:00 · url=https://github.com/UpstageAI/Gym · raw={\"repo\":\"UpstageAI/Gym\",\"parent\":\"NVIDIA-NeMo/Gym\"}"},{"ref":"E52","kind":"event","title":"Experience Stronger Korean Fluency In The New Solar Llms","date":"2026-01-26T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/en/experience-stronger-korean-fluency-in-the-new-solar-llms","signal_url":"https://onlylabs.fyi/signals/05849282-9e27-4730-b487-a9bb9b1e7f2b","signal_json_url":"https://onlylabs.fyi/signals/05849282-9e27-4730-b487-a9bb9b1e7f2b/signal.json","text":"post_published · Experience Stronger Korean Fluency In The New Solar Llms · signal_desk=talking · occurred_at=2026-01-26T00:00:00.000Z · url=https://www.upstage.ai/blog/en/experience-stronger-korean-fluency-in-the-new-solar-llms"},{"ref":"E53","kind":"event","title":"UpstageAI/transformers","date":"2026-01-09T02:49:16+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/transformers","signal_url":"https://onlylabs.fyi/signals/00a08825-977b-40e9-86f7-9d88226a2434","signal_json_url":"https://onlylabs.fyi/signals/00a08825-977b-40e9-86f7-9d88226a2434/signal.json","text":"repo_forked · UpstageAI/transformers · signal_desk=forks · occurred_at=2026-01-09T02:49:16+00:00 · url=https://github.com/UpstageAI/transformers · raw={\"repo\":\"UpstageAI/transformers\",\"parent\":\"huggingface/transformers\"}"},{"ref":"E54","kind":"event","title":"Struggling To Process Loooooooong Document Images With Generative Ai","date":"2025-12-17T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/en/struggling-to-process-loooooooong-document-images-with-generative-ai","signal_url":"https://onlylabs.fyi/signals/40d5c97c-ebb7-4eb7-8bbf-ffc855d65a07","signal_json_url":"https://onlylabs.fyi/signals/40d5c97c-ebb7-4eb7-8bbf-ffc855d65a07/signal.json","text":"post_published · Struggling To Process Loooooooong Document Images With Generative Ai · signal_desk=talking · occurred_at=2025-12-17T00:00:00.000Z · url=https://www.upstage.ai/blog/en/struggling-to-process-loooooooong-document-images-with-generative-ai"},{"ref":"E55","kind":"event","title":"Syn Pro Japan Fm","date":"2025-11-21T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/news/syn-pro-japan-fm","signal_url":"https://onlylabs.fyi/signals/f7284a03-957f-46fe-be75-b6dadefbc52a","signal_json_url":"https://onlylabs.fyi/signals/f7284a03-957f-46fe-be75-b6dadefbc52a/signal.json","text":"post_published · Syn Pro Japan Fm · signal_desk=talking · occurred_at=2025-11-21T00:00:00.000Z · url=https://www.upstage.ai/news/syn-pro-japan-fm"},{"ref":"E56","kind":"event","title":"Extract Structured Data From Any Document Information Extract Api Is Live","date":"2025-10-24T00:00:00.000Z","date_source":"page.visible_date","source_url":"https://www.upstage.ai/blog/en/extract-structured-data-from-any-document--information-extract-api-is-live","signal_url":"https://onlylabs.fyi/signals/1a6588de-4de4-40b3-9022-b5f9bec1c51c","signal_json_url":"https://onlylabs.fyi/signals/1a6588de-4de4-40b3-9022-b5f9bec1c51c/signal.json","text":"post_published · Extract Structured Data From Any Document Information Extract Api Is Live · signal_desk=talking · occurred_at=2025-10-24T00:00:00.000Z · url=https://www.upstage.ai/blog/en/extract-structured-data-from-any-document--information-extract-api-is-live"},{"ref":"E57","kind":"event","title":"UpstageAI/tau2-bench","date":"2025-08-07T05:50:18+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/tau2-bench","signal_url":"https://onlylabs.fyi/signals/8a7d4f11-8520-44a7-94c7-2f6a61bb7e34","signal_json_url":"https://onlylabs.fyi/signals/8a7d4f11-8520-44a7-94c7-2f6a61bb7e34/signal.json","text":"repo_forked · UpstageAI/tau2-bench · signal_desk=forks · occurred_at=2025-08-07T05:50:18+00:00 · url=https://github.com/UpstageAI/tau2-bench · raw={\"repo\":\"UpstageAI/tau2-bench\",\"parent\":\"sierra-research/tau2-bench\"}"},{"ref":"E58","kind":"event","title":"UpstageAI/upstage-open-chat-test","date":"2025-07-03T11:25:10+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/upstage-open-chat-test","signal_url":"https://onlylabs.fyi/signals/f1eed0b6-d608-40c1-b7c4-8d265c81ae53","signal_json_url":"https://onlylabs.fyi/signals/f1eed0b6-d608-40c1-b7c4-8d265c81ae53/signal.json","text":"repo_forked · UpstageAI/upstage-open-chat-test · signal_desk=forks · occurred_at=2025-07-03T11:25:10+00:00 · url=https://github.com/UpstageAI/upstage-open-chat-test · raw={\"repo\":\"UpstageAI/upstage-open-chat-test\",\"parent\":\"UpstageAI/upstage-open-chat\"}"},{"ref":"E59","kind":"event","title":"UpstageAI/solarbox-opentelemetry-collector-contrib","date":"2025-04-22T02:21:32+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/solarbox-opentelemetry-collector-contrib","signal_url":"https://onlylabs.fyi/signals/6711e504-9365-4184-918b-c04c17b60b66","signal_json_url":"https://onlylabs.fyi/signals/6711e504-9365-4184-918b-c04c17b60b66/signal.json","text":"repo_forked · UpstageAI/solarbox-opentelemetry-collector-contrib · signal_desk=forks · occurred_at=2025-04-22T02:21:32+00:00 · url=https://github.com/UpstageAI/solarbox-opentelemetry-collector-contrib · raw={\"repo\":\"UpstageAI/solarbox-opentelemetry-collector-contrib\",\"parent\":\"open-telemetry/opentelemetry-collector-contrib\"}"},{"ref":"E60","kind":"event","title":"UpstageAI/tensorizer","date":"2025-01-17T07:12:34+00:00","date_source":"source","source_url":"https://github.com/UpstageAI/tensorizer","signal_url":"https://onlylabs.fyi/signals/8bc579b2-237e-454c-93f9-75fa94659122","signal_json_url":"https://onlylabs.fyi/signals/8bc579b2-237e-454c-93f9-75fa94659122/signal.json","text":"repo_forked · UpstageAI/tensorizer · signal_desk=forks · occurred_at=2025-01-17T07:12:34+00:00 · url=https://github.com/UpstageAI/tensorizer · raw={\"repo\":\"UpstageAI/tensorizer\",\"parent\":\"coreweave/tensorizer\"}"}]}