{"schema_version":"onlylabs.public_analysis.v1","url":"https://onlylabs.fyi/analysis/reka","json_url":"https://onlylabs.fyi/analysis/reka/analysis.json","evidence_json_url":"https://onlylabs.fyi/analysis/reka/evidence.json","generated_at":"2026-06-28T02:17:46.556Z","analysis":{"org_slug":"reka","url":"https://onlylabs.fyi/analysis/reka","json_url":"https://onlylabs.fyi/analysis/reka/analysis.json","evidence_json_url":"https://onlylabs.fyi/analysis/reka/evidence.json","dossier_url":"https://onlylabs.fyi/labs/reka","org":{"slug":"reka","name":"Reka AI","category":"neolab","category_label":"Neolab","homepage_url":"https://reka.ai"},"title":"Reka AI analysis","summary":"Reka is evolving from a multimodal model builder into a physical-AI company. The evidence shows a lab that ships compact, deployment-flexible vision-language models (Flash at 21B, Edge at 7B), layers enterprise agentic platforms on top (Nexus, Vision, Research), and is now orienting research toward world models, embodied data, and physical reasoning. The $110M raise backed by NVIDIA and Snowflake, the Moonvalley…","markdown":"## Thesis\n\nReka is evolving from a multimodal model builder into a physical-AI company. The evidence shows a lab that ships compact, deployment-flexible vision-language models (Flash at 21B, Edge at 7B) [P11](https://reka.ai/news/reka-flash-efficient-and-capable-multimodal-language-models)[P14](https://reka.ai/news/introducing-reka-flash)[P19](https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai), layers enterprise agentic platforms on top (Nexus, Vision, Research) [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P9](https://reka.ai/news/introducing-nexus)[P12](https://reka.ai/news/reka-vision-intelligence-made-visible), and is now orienting research toward world models, embodied data, and physical reasoning [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models)[E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f). The $110M raise backed by NVIDIA and Snowflake [P23](https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms), the Moonvalley team acquisition [W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html)[W2](https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html), and the Oracle defense ecosystem entry [P1](https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness) all point to a scaling phase focused on real-world deployment rather than pure benchmark competition. Reka’s differentiator is token-efficient multimodal architectures that run on-prem, on-device, or at the edge — a positioning that directly targets enterprises and public-sector buyers with data-sovereignty requirements [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P10](https://reka.ai/news/multimodal-with-reka-mongodb)[P25](https://reka.ai/news/how-reka-vision-is-powering-the-future-of-ai-driven-security).\n\n## Signal desks\n\n### Hiring\n\n- **Repeated hiring themes:** Robotics Research Lead [E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f), GPU Performance Engineer [E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d), Data Intelligence [E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d), Applied AI [E49](https://jobs.ashbyhq.com/reka/e059e019-b7f1-4ccb-a53b-1db4f821621c), ML Engineer [E8](https://jobs.ashbyhq.com/reka/87ec3e1d-3083-4dc8-9167-89167dc683e1), Product Engineer [E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652), and generalist ML [E60](https://jobs.ashbyhq.com/reka/731e2ebc-2e66-458a-9962-0bc880016941). The robotics and GPU roles are new since early 2026 and signal a physical-AI and inference-optimization buildout.\n- **Locations/hubs:** Roles span US, UK, and Singapore; all are remote-first [E8](https://jobs.ashbyhq.com/reka/87ec3e1d-3083-4dc8-9167-89167dc683e1)[E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f)[E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652)[E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d)[E49](https://jobs.ashbyhq.com/reka/e059e019-b7f1-4ccb-a53b-1db4f821621c)[E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d)[E60](https://jobs.ashbyhq.com/reka/731e2ebc-2e66-458a-9962-0bc880016941). Singapore has appeared only recently (Data Intelligence, GPU Performance Engineer), suggesting an Asia-Pacific expansion or talent diversification play [E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d)[E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d).\n- **Implied infrastructure spend:** GPU Performance Engineer implies custom kernel work and serving optimization for on-prem/edge deployments [E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d). Data Intelligence implies investment in multimodal data pipelines and synthetic data generation at scale [E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d). Robotics Research Lead implies sim-to-real infrastructure and embodied evaluation environments [E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f).\n- **Team composition:** The founding team and many hires come from DeepMind and FAIR [W3](https://www.gravityer.com/jobs/member-of-technical-staff-applied-ai-reka). Hiring spans Research, AI Engineering, and Technical teams, with product engineering indicating GTM buildout [E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652).\n\n### Forks\n\n- **Agent/integration frameworks:** llama_index [E37](https://github.com/reka-ai/llama_index) and langchain [E41](https://github.com/reka-ai/langchain) were forked in mid-2024, consistent with early RAG and agent-integration work that preceded the Nexus platform launch.\n- **Video and 3D research:** Video-Depth-Anything [E58](https://github.com/reka-ai/Video-Depth-Anything), TAPIP3D [E57](https://github.com/reka-ai/TAPIP3D), and insert-anything [E56](https://github.com/reka-ai/insert-anything) were forked in mid-2025, aligned with the Reka Vision platform release and video-understanding research. The diffusers fork [E59](https://github.com/reka-ai/diffusers) suggests exploration of generative visual models.\n- **Low-profile signals:** The Archon fork [E40](https://github.com/reka-ai/Archon) (agent architecture) in late 2024 is consistent with the agentic-era narrative Reka began pushing publicly in early 2025 [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now). Fork activity overall is modest and appears more exploratory than indicative of a major upstream dependency strategy.\n\n### Releases\n\n- **Model releases:** Reka Flash 3 (21B, Apache 2.0, 122 downloads on HF, 391 likes) [E1](https://huggingface.co/RekaAI/reka-flash-3); Reka Flash 3.1 (21B, Apache 2.0, 45 downloads, 100 likes) [E3](https://huggingface.co/RekaAI/reka-flash-3.1); Reka Edge 2603 (7B, image-text-to-text, 399 downloads, 131 likes) [E2](https://huggingface.co/RekaAI/reka-edge-2603). Downloads are modest but Edge shows the strongest per-unit traction, consistent with the edge-deployment narrative.\n- **Infrastructure tools:** cs2-dem-renderer (Go, for converting CS2 demo files to video + annotations) was released with multiple version tags in June 2026 [E5](https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.6.5)[E6](https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.2.8)[E7](https://github.com/reka-ai/cs2-dem-renderer); vllm-reka (vLLM plugin for Reka models) released February 2026 with multiple version bumps through April [E22](https://github.com/reka-ai/vllm-reka/releases/tag/v0.17.0)[E23](https://github.com/reka-ai/vllm-reka/releases/tag/v0.15.0)[E32](https://github.com/reka-ai/vllm-reka); Reka Quant (quantization library) released July 2025 [E28](https://github.com/reka-ai/rekaquant).\n- **Developer tooling:** n8n-nodes-reka has seen 10+ version tags from January to May 2026 [E17](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.7.1)[E18](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.7.0)[E20](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.6.4)[E24](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.6.0)[E46](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.5.0)[E47](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.5)[E48](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.4)[E50](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.4-beta.3)[E51](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.4.3)[E52](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.4.2), indicating sustained investment in no-code/low-code workflow integration. Reka MCP server released April 2026 [E19](https://github.com/reka-ai/reka-mcp). SDKs exist for Python, .NET, and TypeScript [E29](https://github.com/reka-ai/reka-sdk-python)[E31](https://github.com/reka-ai/api-examples-dotnet)[E36](https://github.com/reka-ai/api-examples-typescript)[E54](https://github.com/reka-ai/reka-sdk-dotnet/releases/tag/v0.1.4-beta).\n- **Benchmarks/datasets:** Vibe-Eval (multimodal chat benchmark, 187 stars) [E26](https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1)[E27](https://github.com/reka-ai/reka-vibe-eval); Research-Eval (search-augmented LLM benchmark, 17 stars) [E30](https://github.com/reka-ai/research-eval); CS2-10k dataset [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[E4](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset); PhysicalRealismBench-U [P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models)[E9](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models).\n\n### Talking\n\n- **Physical AI and world models:** The dominant emerging theme in mid-2026. CS2-10k dataset announcement frames Counter-Strike as a substrate for world-model research [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[E4](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset). PhysicalRealismBench-U argues current VLMs fail at intuitive physics [P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models)[E9](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models). The Moonvalley partnership press release explicitly positions Reka as “building foundational intelligence for the physical world” [W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html)[W2](https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html)[E10](https://reka.ai/news/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai).\n- **Multimodal vision applications:** Heavy public writing about Reka Vision for security (false-alarm reduction, natural-language search) [P4](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai)[P25](https://reka.ai/news/how-reka-vision-is-powering-the-future-of-ai-driven-security)[P21](https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works)[E13](https://reka.ai/news/reka-vision-intelligence-made-visible)[E15](https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works)[E55](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai), broadcast archive monetization (Shutterstock partnership) [P6](https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive)[P18](https://reka.ai/news/reka-announces-partnership-with-shutterstock)[E14](https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive), and smart-home benchmarks [P4](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai)[E55](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai). These posts serve as GTM content for enterprise buyers.\n- **Agentic enterprise narrative:** Multiple posts frame Nexus as an “AI workforce” for automating multimodal unstructured data processing [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P9](https://reka.ai/news/introducing-nexus)[P13](https://reka.ai/news/reka-launches-nexus). The messaging targets CIOs/CTOs with private-deployment and transparency angles.\n- **Deployment flexibility and sovereign AI:** Posts emphasize on-prem, on-device, and private-cloud deployment [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P10](https://reka.ai/news/multimodal-with-reka-mongodb)[P14](https://reka.ai/news/introducing-reka-flash)[P22](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter), the OpenRouter integration [P22](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter)[E21](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter), and public-sector private AI models [E25](https://reka.ai/news/trust-and-intelligence-why-the-public-sector-is-moving-toward-private-ai-models). Oracle defense ecosystem entry reinforces this narrative [P1](https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness).\n- **Research transparency:** Reka publishes model cards, technical reports, and open-source releases (Flash 3, Flash 3.1, Reka Quant, Vibe-Eval, Research-Eval) [P14](https://reka.ai/news/introducing-reka-flash)[P28](https://reka.ai/news/reka-flash-3-1-and-reka-quant)[P26](https://reka.ai/news/vibe-eval)[P2](https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms)[E1](https://huggingface.co/RekaAI/reka-flash-3)[E3](https://huggingface.co/RekaAI/reka-flash-3.1)[E28](https://github.com/reka-ai/rekaquant), maintaining credibility with the research community.\n- **Traction on posts:** HN traction is minimal (CS2-10k at 2 points [E4](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset), Reka Vision at 5 points [E13](https://reka.ai/news/reka-vision-intelligence-made-visible), RL post at 1 point [E26](https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1)), indicating limited organic developer-mindshare relative to larger labs.\n\n## Shipping\n\nReka ships across four lanes: **(1) Models** — Flash 3, Flash 3.1, and Edge are on Hugging Face under permissive or custom licenses [E1](https://huggingface.co/RekaAI/reka-flash-3)[E2](https://huggingface.co/RekaAI/reka-edge-2603)[E3](https://huggingface.co/RekaAI/reka-flash-3.1); Core was announced April 2024 as a frontier-class model but is not open-weight [P16](https://reka.ai/news/reka-core-our-frontier-class-multimodal-language-model). **(2) Platforms** — Nexus (enterprise AI workers), Reka Vision (video/image understanding and search), and Reka Research (search-augmented LLM) form the commercial product layer [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P9](https://reka.ai/news/introducing-nexus)[P12](https://reka.ai/news/reka-vision-intelligence-made-visible)[P13](https://reka.ai/news/reka-launches-nexus)[P2](https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms). **(3) Developer tooling** — n8n community node, MCP server, vLLM plugin, Reka Quant, CLI tools, and multi-language SDKs [P3](https://reka.ai/news/introducing-the-reka-n8n-community-node-ai-powered-video-and-image-analysis-made-easy)[E17](https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.7.1)[E19](https://github.com/reka-ai/reka-mcp)[E22](https://github.com/reka-ai/vllm-reka/releases/tag/v0.17.0)[E28](https://github.com/reka-ai/rekaquant)[E29](https://github.com/reka-ai/reka-sdk-python)[E38](https://github.com/reka-ai/terminal-tools)[E44](https://github.com/reka-ai/reka-cli). **(4) Research artifacts** — CS2-10k, PhysicalRealismBench-U, Vibe-Eval, Research-Eval, and the cs2-dem-renderer pipeline [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models)[P26](https://reka.ai/news/vibe-eval)[P2](https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms)[E5](https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.6.5).\n\n## Research themes\n\n1. **Physical intelligence and world models:** The dominant 2026 research thrust. CS2-10k provides 10,000+ hours of egocentric video with dense action annotations for training interactive world models [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset). PhysicalRealismBench-U probes VLM failures on object permanence, motion, and collision — tasks that “even kids would easily solve” [P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models). The Moonvalley acquisition adds dedicated researchers to this track [W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html)[W2](https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html).\n2. **Token-efficient multimodal architectures:** Reka Edge uses a ConvNeXt V2 vision encoder producing only 64 tokens per image tile, paired with a 6.4B transformer trained from scratch [P19](https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai). This architectural choice directly targets latency and cost for edge deployment.\n3. **Reinforcement learning for reasoning:** Reka Flash 3.1 improvements came from “major upgrades to our RL stack, including a new RL algorithm and significant scalability improvements” using RLOO with model-based and rule-based rewards [P28](https://reka.ai/news/reka-flash-3-1-and-reka-quant)[P14](https://reka.ai/news/introducing-reka-flash)[E26](https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1).\n4. **Search-augmented LLM evaluation:** Research-Eval addresses a gap between saturated benchmarks (SimpleQA) and artificial puzzle benchmarks (BrowseComp), with 374 diverse questions and checklist-based LLM judging [P2](https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms)[E30](https://github.com/reka-ai/research-eval).\n5. **Quantization for deployment:** Reka Quant achieves near-lossless 3.5-bit quantization (1.6 avg performance degradation vs. 6.8 for standard Q3_K_S), enabling model serving in resource-constrained settings [P28](https://reka.ai/news/reka-flash-3-1-and-reka-quant)[E28](https://github.com/reka-ai/rekaquant).\n\n## Hiring & scaling\n\nActive hiring as of mid-2026 spans seven distinct roles across three continents [E8](https://jobs.ashbyhq.com/reka/87ec3e1d-3083-4dc8-9167-89167dc683e1)[E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f)[E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652)[E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d)[E49](https://jobs.ashbyhq.com/reka/e059e019-b7f1-4ccb-a53b-1db4f821621c)[E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d)[E60](https://jobs.ashbyhq.com/reka/731e2ebc-2e66-458a-9962-0bc880016941). The most strategically telling roles:\n\n- **Robotics Research Lead (UK Remote)** — opened May 2026, signals a new physical-AI/embodied research team that did not previously exist at Reka [E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f).\n- **GPU Performance Engineer (US/UK/Singapore)** — opened January 2026, implies custom CUDA/kernel work for serving models on-prem and at the edge, not just API hosting [E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d).\n- **Data Intelligence (US/UK/Singapore)** — opened May 2026, sits on the Research team and implies investment in multimodal data curation, synthetic data generation, and data quality pipelines [E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d).\n- **Applied AI and Product Engineer** — both opened early 2026, signal commercialization and GTM buildout [E49](https://jobs.ashbyhq.com/reka/e059e019-b7f1-4ccb-a53b-1db4f821621c)[E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652).\n\nThe Moonvalley team acquisition in June 2026 adds an unspecified number of AI researchers and engineers [W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html)[W2](https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html). Headcount trajectory, combined with the $110M raise [P23](https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms), suggests Reka is in a deliberate scaling phase after an extended period of capital-efficient model development.\n\n## Category implications\n\n- **Physical AI / World Models:** CS2-10k and PhysicalRealismBench-U [P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models) plus the robotics hire [E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f) and Moonvalley acquisition [W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html) signal that Reka sees world models as the next frontier beyond language-and-vision chatbots. This puts them in direct research competition with labs investing in video generation and embodied AI, though Reka’s approach is data-centric (game engine data) rather than hardware-centric.\n- **Enterprise multimodal platforms:** Nexus, Vision, and Research [P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now)[P9](https://reka.ai/news/introducing-nexus)[P12](https://reka.ai/news/reka-vision-intelligence-made-visible) are positioned as private-deployable alternatives to API-only agent platforms. The Oracle defense [P1](https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness), Shutterstock [P18](https://reka.ai/news/reka-announces-partnership-with-shutterstock), and Wyze/SmartHome-Bench [P4](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai) relationships suggest GTM focus on security, media, and public-sector verticals where data sovereignty matters.\n- **Edge/deployed inference:** Reka Edge (7B, ConvNeXt encoder) [P19](https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai), the GPU Performance Engineer hire [E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d), Reka Quant [P28](https://reka.ai/news/reka-flash-3-1-and-reka-quant), and the OpenRouter integration [P22](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter) all point to a strategy of making models run anywhere — on-device, on-prem, or via managed API. This is a cost-of-ownership play against larger models that require cloud API calls.\n- **Research infrastructure:** The cs2-dem-renderer [E7](https://github.com/reka-ai/cs2-dem-renderer), vllm-reka [E32](https://github.com/reka-ai/vllm-reka), and Reka Quant [E28](https://github.com/reka-ai/rekaquant) releases suggest Reka builds its own training and inference tooling rather than relying entirely on external frameworks. This could reduce dependency on ecosystem partners but also increases engineering burden.\n- **Hiring implications:** The Singapore location appearing in recent roles [E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d)[E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d) may signal either an Asia-Pacific commercial expansion (consistent with the Telkom Indonesia collaboration [P24](https://reka.ai/news/reka-collabs-with-telkom)) or access to GPU engineering talent. The remote-first posture with US/UK hubs suggests Reka competes for the same talent pool as DeepMind, Anthropic, and FAIR [W3](https://www.gravityer.com/jobs/member-of-technical-staff-applied-ai-reka).\n\n## Traction highlights\n\n- **Funding:** $110M raised July 2025 from NVIDIA and Snowflake [P23](https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms).\n- **Enterprise partnerships:** Shutterstock (multi-year, data licensing + customer) [P18](https://reka.ai/news/reka-announces-partnership-with-shutterstock); Oracle Defense Ecosystem [P1](https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness); Telkom Indonesia (MoU for Bahasa Indonesia models) [P24](https://reka.ai/news/reka-collabs-with-telkom); Wyze (SmartHome-Bench evaluation partner) [P4](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai); Turing Video (cited as a Reka Vision customer) [P23](https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms).\n- **Recognition:** Tracxn Emerging Award in AI Infrastructure category, October 2025 [P27](https://reka.ai/news/from-research-to-real-world-impact-reka-recognized-in-tracxn-s-emerging-awards-for-ai-infrastructure).\n- **Model adoption:** Reka Flash 3 has 391 likes on Hugging Face [E1](https://huggingface.co/RekaAI/reka-flash-3); Reka Edge 2603 has 399 downloads and 131 likes [E2](https://huggingface.co/RekaAI/reka-edge-2603). OpenRouter integration went live April 2026 [P22](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter).\n- **Developer ecosystem:** n8n community node (10+ releases) , MCP servers [E19](https://github.com/reka-ai/reka-mcp)[E34](https://github.com/reka-ai/reka-research-mcp), vLLM plugin [E32](https://github.com/reka-ai/vllm-reka), multi-language SDKs [E29](https://github.com/reka-ai/reka-sdk-python)[E31](https://github.com/reka-ai/api-examples-dotnet)[E36](https://github.com/reka-ai/api-examples-typescript), and a free API tier with $10/month credits [P17](https://reka.ai/news/end-of-summer-updates).\n- **HN/community buzz:** Low. The highest HN traction is 5 points on the Reka Vision launch [E13](https://reka.ai/news/reka-vision-intelligence-made-visible); most posts have 0–2 points [E4](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset)[E26](https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1). This suggests Reka has not yet captured significant organic developer attention despite substantial shipping volume.\n\n## Sources\n\n[P1](https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness) Reka Joins Oracle’s Defense Ecosystem (Oct 2025)\n[P2](https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms) Introducing Research-Eval (Aug 2025)\n[P3](https://reka.ai/news/introducing-the-reka-n8n-community-node-ai-powered-video-and-image-analysis-made-easy) Introducing the Reka n8n Community Node (Feb 2026)\n[P4](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai) Reka Vision Sets New Standard for Smart Home & Security AI (Nov 2025)\n[P5](https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now) Entering the Agentic Era: Why Businesses Need an AI Workforce Now (Mar 2025)\n[P6](https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive) Your Archive Is a Revenue Engine (May 2026)\n[P7](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset) CS2-10k: A Large-Scale Egocentric Counter-Strike 2 Dataset (Jun 2026)\n[P8](https://reka.ai/news/reka-flash-updates) Reka Flash Updates (Oct 2024)\n[P9](https://reka.ai/news/introducing-nexus) Introducing Nexus (Mar 2025)\n[P10](https://reka.ai/news/multimodal-with-reka-mongodb) Multimodal with Reka & MongoDB (Sep 2024)\n[P11](https://reka.ai/news/reka-flash-efficient-and-capable-multimodal-language-models) Reka Flash: Efficient and Capable Multimodal Language Models (Feb 2024)\n[P12](https://reka.ai/news/reka-vision-intelligence-made-visible) Reka Vision: Intelligence Made Visible (Jul 2025)\n[P13](https://reka.ai/news/reka-launches-nexus) Reka Launches Nexus (Mar 2025)\n[P14](https://reka.ai/news/introducing-reka-flash) Introducing Reka Flash 3 (Mar 2025)\n[P15](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models) PhysicalRealismBench-U (Jun 2026)\n[P16](https://reka.ai/news/reka-core-our-frontier-class-multimodal-language-model) Reka Core: Our Frontier Class Multimodal Language Model (Apr 2024)\n[P17](https://reka.ai/news/end-of-summer-updates) End of Summer Updates (Sep 2025)\n[P18](https://reka.ai/news/reka-announces-partnership-with-shutterstock) Reka Announces Partnership with Shutterstock (Jun 2024)\n[P19](https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai) Reka Edge: Frontier-Level Edge Intelligence for Physical AI (Mar 2026)\n[P20](https://reka.ai/news/how-to-leverage-reka-research-to-build-smarter-ai-apps) How to Leverage Reka Research to Build Smarter AI Apps (Sep 2025)\n[P21](https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works) Adding Reka Vision Without Replacing VMS (May 2026)\n[P22](https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter) Switch Models, Zero Code Changes: Reka Edge Now Available on OpenRouter (Apr 2026)\n[P23](https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms) Reka Secures $110 Million (Jul 2025)\n[P24](https://reka.ai/news/reka-collabs-with-telkom) Reka Collabs with Telkom (Aug 2024)\n[P25](https://reka.ai/news/how-reka-vision-is-powering-the-future-of-ai-driven-security) How Reka Vision Is Powering the Future of AI-Driven Security (Jul 2025)\n[P26](https://reka.ai/news/vibe-eval) Vibe-Eval (May 2024)\n[P27](https://reka.ai/news/from-research-to-real-world-impact-reka-recognized-in-tracxn-s-emerging-awards-for-ai-infrastructure) Reka Recognized in Tracxn’s Emerging Awards for AI Infrastructure (Oct 2025)\n[P28](https://reka.ai/news/reka-flash-3-1-and-reka-quant) Reka Flash 3.1 and Reka Quant (Jul 2025)\n[E1](https://huggingface.co/RekaAI/reka-flash-3) Reka Flash 3 on Hugging Face (Mar 2025)\n[E2](https://huggingface.co/RekaAI/reka-edge-2603) Reka Edge 2603 on Hugging Face (Mar 2026)\n[E3](https://huggingface.co/RekaAI/reka-flash-3.1) Reka Flash 3.1 on Hugging Face (May 2025)\n[E4](https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset) CS2-10k post on HN (Jun 2026)\n[E5](https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.6.5) cs2-dem-renderer v1.41.6.5 (Jun 2026)\n[E6](https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.2.8) cs2-dem-renderer v1.41.2.8 (Jun 2026)\n[E7](https://github.com/reka-ai/cs2-dem-renderer) cs2-dem-renderer repo (Jun 2026)\n[E8](https://jobs.ashbyhq.com/reka/87ec3e1d-3083-4dc8-9167-89167dc683e1) MTS Machine Learning Engineer job (Jun 2026)\n[E9](https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models) PhysicalRealismBench post (Jun 2026)\n[E10](https://reka.ai/news/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai) Reka and Moonvalley join forces post (Jun 2026)\n[E11](https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f) MTS Robotics Research Lead job (May 2026)\n[E12](https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652) MTS Product Engineer job (May 2026)\n[E13](https://reka.ai/news/reka-vision-intelligence-made-visible) Reka Vision launch HN (Jul 2025)\n[E14](https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive) Archive Revenue Engine post (May 2026)\n[E15](https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works) Adding Reka Vision Without Replacing VMS post (May 2026)\n[E16](https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d) MTS Data Intelligence job (May 2026)\n n8n-nodes-reka releases (Jan–May 2026)\n vllm-reka releases (Apr 2026)\n[E25](https://reka.ai/news/trust-and-intelligence-why-the-public-sector-is-moving-toward-private-ai-models) Trust and Intelligence: Public Sector Private AI post (Apr 2026)\n[E26](https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1) RL for Reka Flash 3.1 post (Jul 2025)\n[E27](https://github.com/reka-ai/reka-vibe-eval) Vibe-Eval repo (May 2024)\n[E28](https://github.com/reka-ai/rekaquant) Reka Quant repo (Jul 2025)\n[E29](https://github.com/reka-ai/reka-sdk-python) reka-sdk-python repo (May 2024)\n[E30](https://github.com/reka-ai/research-eval) Research-Eval repo (Aug 2025)\n[E31](https://github.com/reka-ai/api-examples-dotnet) api-examples-dotnet repo (Aug 2025)\n[E32](https://github.com/reka-ai/vllm-reka) vllm-reka repo (Feb 2026)\n[E33](https://github.com/reka-ai/api-examples-python) api-examples-python repo (Jul 2025)\n[E34](https://github.com/reka-ai/reka-research-mcp) reka-research-mcp repo (Jul 2025)\n[E35](https://github.com/reka-ai/clip-api-examples) clip-api-examples repo (Feb 2026)\n[E36](https://github.com/reka-ai/api-examples-typescript) api-examples-typescript repo (Aug 2025)\n[E37](https://github.com/reka-ai/llama_index) llama_index fork (Aug 2024)\n[E38](https://github.com/reka-ai/terminal-tools) terminal-tools repo (Nov 2025)\n[E39](https://github.com/reka-ai/n8n-nodes-reka) n8n-nodes-reka repo (Jan 2026)\n[E40](https://github.com/reka-ai/Archon) Archon fork (Oct 2024)\n[E41](https://github.com/reka-ai/langchain) langchain fork (Aug 2024)\n[E42](https://github.com/reka-ai/colab) colab repo (Sep 2023)\n[E43](https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai) Reka Edge for Physical AI post (Mar 2026)\n[E44](https://github.com/reka-ai/reka-cli) reka-cli repo (Mar 2026)\n[E45](https://reka.ai/news/introducing-the-reka-n8n-community-node-ai-powered-video-and-image-analysis-made-easy) n8n Community Node post (Feb 2026)\n n8n-nodes-reka releases (Jan–Feb 2026)\n[E53](https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d) MTS GPU Performance Engineer job (Jan 2026)\n[E54](https://github.com/reka-ai/reka-sdk-dotnet/releases/tag/v0.1.4-beta) reka-sdk-dotnet release (Jan 2026)\n[E55](https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai) Reka Vision Smart Home Security post (Nov 2025)\n[E56](https://github.com/reka-ai/insert-anything) insert-anything fork (Aug 2025)\n[E57](https://github.com/reka-ai/TAPIP3D) TAPIP3D fork (Jun 2025)\n[E58](https://github.com/reka-ai/Video-Depth-Anything) Video-Depth-Anything fork (Jun 2025)\n[E59](https://github.com/reka-ai/diffusers) diffusers fork (Dec 2024)\n[E60](https://jobs.ashbyhq.com/reka/731e2ebc-2e66-458a-9962-0bc880016941) MTS ML job (Aug 2023)\n[W1](https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html) Moonvalley press release (Jun 2026)\n[W2](https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html) Moonvalley APAC press release (Jun 2026)\n[W3](https://www.gravityer.com/jobs/member-of-technical-staff-applied-ai-reka) Applied AI job description via Gravity Engineering (Jun 2026)\n[W4](https://reka.ai/labs) Reka Labs page (accessed Jun 2026)","generated_at":"2026-06-27T19:36:47.442+00:00","citations":[{"url":"https://reka.ai/news/reka-flash-efficient-and-capable-multimodal-language-models","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/introducing-reka-flash","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/entering-the-agentic-era-why-businesses-need-an-ai-workforce-now","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/introducing-nexus","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-vision-intelligence-made-visible","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/9cdfc4f4-9594-49eb-9eaa-08aff01aa22f","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://reka.ai/news/reka-secures-110-million-to-accelerate-adoption-of-its-multimodal-ai-platforms","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://www.prnewswire.com/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797869.html","path":null,"label":"prnewswire.com/news-releases","type":"external"},{"url":"https://www.prnewswire.com/apac/news-releases/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai-302797873.html","path":null,"label":"prnewswire.com/apac","type":"external"},{"url":"https://reka.ai/news/reka-joins-oracle-s-defense-ecosystem-to-advance-multimodal-ai-for-mission-readiness","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/multimodal-with-reka-mongodb","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/how-reka-vision-is-powering-the-future-of-ai-driven-security","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/6595a48b-8dfc-479d-ba2a-0bf9d188ec4d","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/32f1d782-9a49-4a2e-8842-ac90d002bc8d","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/e059e019-b7f1-4ccb-a53b-1db4f821621c","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/87ec3e1d-3083-4dc8-9167-89167dc683e1","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/4676af32-41e8-428e-af64-63c08e574652","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://jobs.ashbyhq.com/reka/731e2ebc-2e66-458a-9962-0bc880016941","path":null,"label":"jobs.ashbyhq.com/reka","type":"external"},{"url":"https://www.gravityer.com/jobs/member-of-technical-staff-applied-ai-reka","path":null,"label":"gravityer.com/jobs","type":"external"},{"url":"https://github.com/reka-ai/llama_index","path":null,"label":"reka-ai/llama_index","type":"external"},{"url":"https://github.com/reka-ai/langchain","path":null,"label":"reka-ai/langchain","type":"external"},{"url":"https://github.com/reka-ai/Video-Depth-Anything","path":null,"label":"reka-ai/Video-Depth-Anything","type":"external"},{"url":"https://github.com/reka-ai/TAPIP3D","path":null,"label":"reka-ai/TAPIP3D","type":"external"},{"url":"https://github.com/reka-ai/insert-anything","path":null,"label":"reka-ai/insert-anything","type":"external"},{"url":"https://github.com/reka-ai/diffusers","path":null,"label":"reka-ai/diffusers","type":"external"},{"url":"https://github.com/reka-ai/Archon","path":null,"label":"reka-ai/Archon","type":"external"},{"url":"https://huggingface.co/RekaAI/reka-flash-3","path":null,"label":"RekaAI/reka-flash-3","type":"external"},{"url":"https://huggingface.co/RekaAI/reka-flash-3.1","path":null,"label":"RekaAI/reka-flash-3.1","type":"external"},{"url":"https://huggingface.co/RekaAI/reka-edge-2603","path":null,"label":"RekaAI/reka-edge-2603","type":"external"},{"url":"https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.6.5","path":null,"label":"reka-ai/cs2-dem-renderer","type":"external"},{"url":"https://github.com/reka-ai/cs2-dem-renderer/releases/tag/1.41.2.8","path":null,"label":"reka-ai/cs2-dem-renderer","type":"external"},{"url":"https://github.com/reka-ai/cs2-dem-renderer","path":null,"label":"reka-ai/cs2-dem-renderer","type":"external"},{"url":"https://github.com/reka-ai/vllm-reka/releases/tag/v0.17.0","path":null,"label":"reka-ai/vllm-reka","type":"external"},{"url":"https://github.com/reka-ai/vllm-reka/releases/tag/v0.15.0","path":null,"label":"reka-ai/vllm-reka","type":"external"},{"url":"https://github.com/reka-ai/vllm-reka","path":null,"label":"reka-ai/vllm-reka","type":"external"},{"url":"https://github.com/reka-ai/rekaquant","path":null,"label":"reka-ai/rekaquant","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.7.1","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.7.0","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.6.4","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.6.0","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.5.0","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.5","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.4","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/v0.4.4-beta.3","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.4.3","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka/releases/tag/0.4.2","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/reka-mcp","path":null,"label":"reka-ai/reka-mcp","type":"external"},{"url":"https://github.com/reka-ai/reka-sdk-python","path":null,"label":"reka-ai/reka-sdk-python","type":"external"},{"url":"https://github.com/reka-ai/api-examples-dotnet","path":null,"label":"reka-ai/api-examples-dotnet","type":"external"},{"url":"https://github.com/reka-ai/api-examples-typescript","path":null,"label":"reka-ai/api-examples-typescript","type":"external"},{"url":"https://github.com/reka-ai/reka-sdk-dotnet/releases/tag/v0.1.4-beta","path":null,"label":"reka-ai/reka-sdk-dotnet","type":"external"},{"url":"https://reka.ai/news/reinforcement-learning-for-reka-flash-3-1","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://github.com/reka-ai/reka-vibe-eval","path":null,"label":"reka-ai/reka-vibe-eval","type":"external"},{"url":"https://github.com/reka-ai/research-eval","path":null,"label":"reka-ai/research-eval","type":"external"},{"url":"https://reka.ai/news/cs2-10k-a-large-scale-egocentric-counter-strike-2-dataset","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/physicalrealismbench-attributable-physical-realism-evaluation-for-video-world-models","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-and-moonvalley-join-forces-to-advance-models-and-infrastructure-for-physical-ai","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-vision-intelligence-made-visible","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/adding-reka-vision-without-replacing-vms-what-actually-works","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-vision-sets-new-standard-for-smart-home-security-ai","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-announces-partnership-with-shutterstock","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/your-archive-is-a-revenue-engine.-most-broadcasters-are-still-treating-it-like-a-hard-drive","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-launches-nexus","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/switch-models-zero-code-changes-reka-edge-now-available-on-openrouter","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/trust-and-intelligence-why-the-public-sector-is-moving-toward-private-ai-models","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-flash-3-1-and-reka-quant","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/vibe-eval","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/introducing-research-eval-a-benchmark-for-search-augmented-llms","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-core-our-frontier-class-multimodal-language-model","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/introducing-the-reka-n8n-community-node-ai-powered-video-and-image-analysis-made-easy","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://github.com/reka-ai/terminal-tools","path":null,"label":"reka-ai/terminal-tools","type":"external"},{"url":"https://github.com/reka-ai/reka-cli","path":null,"label":"reka-ai/reka-cli","type":"external"},{"url":"https://reka.ai/news/reka-collabs-with-telkom","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/from-research-to-real-world-impact-reka-recognized-in-tracxn-s-emerging-awards-for-ai-infrastructure","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://github.com/reka-ai/reka-research-mcp","path":null,"label":"reka-ai/reka-research-mcp","type":"external"},{"url":"https://reka.ai/news/end-of-summer-updates","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/reka-flash-updates","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/how-to-leverage-reka-research-to-build-smarter-ai-apps","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://github.com/reka-ai/api-examples-python","path":null,"label":"reka-ai/api-examples-python","type":"external"},{"url":"https://github.com/reka-ai/clip-api-examples","path":null,"label":"reka-ai/clip-api-examples","type":"external"},{"url":"https://github.com/reka-ai/n8n-nodes-reka","path":null,"label":"reka-ai/n8n-nodes-reka","type":"external"},{"url":"https://github.com/reka-ai/colab","path":null,"label":"reka-ai/colab","type":"external"},{"url":"https://reka.ai/news/reka-edge-frontier-level-edge-intelligence-for-physical-ai","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/news/introducing-the-reka-n8n-community-node-ai-powered-video-and-image-analysis-made-easy","path":null,"label":"reka.ai/news","type":"external"},{"url":"https://reka.ai/labs","path":null,"label":"reka.ai/labs","type":"external"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-pro","workflow":"onlylabs-deepagents-analysis-v3","agent":"deepagents"},"evidence":{"total":92,"pages":28,"events":105,"web":4,"signal_desks":{"forks":7,"repos":16,"hiring":7,"talking":12,"releases":18},"data_radar_lanes":null,"data_radar_matches":null}},"signal_counts":{"total":105,"model_released":3,"release":25,"repo_new":17,"repo_forked":7,"post_published":46,"job_opened":7}}