{"schema_version":"onlylabs.public_signal.v1","title":"Mistral AI Writing: Heaps do lie: debugging a memory leak in vLLM.","description":"Mistral AI writing signal with public source context, captured evidence pages, related signals, and data-business radar classification.","url":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9","json_url":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9/signal.json","generated_at":"2026-06-27T00:36:53.985Z","evidence_latest_fetched_at":"2026-06-26T20:23:09.659613+00:00","signal_first_seen_at":"2026-06-26T20:22:13.235771+00:00","org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab","category_label":"Frontier lab","dossier_url":"https://onlylabs.fyi/labs/mistral","dossier_json_url":"https://onlylabs.fyi/labs/mistral/dossier.json"},"related_urls":{"signal":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9","signal_json":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9/signal.json","source":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","lab_dossier":"https://onlylabs.fyi/labs/mistral","lab_dossier_json":"https://onlylabs.fyi/labs/mistral/dossier.json","analysis":"https://onlylabs.fyi/analysis/mistral","analysis_json":"https://onlylabs.fyi/analysis/mistral/analysis.json","analysis_evidence_json":"https://onlylabs.fyi/analysis/mistral/evidence.json","category":"https://onlylabs.fyi/frontier","category_json":"https://onlylabs.fyi/frontier.json","category_feed":"https://onlylabs.fyi/frontier/feed.xml","category_signals_json":"https://onlylabs.fyi/signals.json","topic":"https://onlylabs.fyi/topics/talking","topic_signals_json":"https://onlylabs.fyi/topics/talking/signals.json","topic_feed":"https://onlylabs.fyi/topics/talking/feed.xml","data_business":null},"answer_pack":{"answer":"Mistral AI published Heaps do lie: debugging a memory leak in vLLM.. This talking signal gives public context for research themes, product direction, policy, or launch framing. High-signal details: Substantive debugging post with low traction · Heaps do lie: debugging a memory leak in vLLM. | Mistral AI Engineering Heaps do lie: debugging a memory leak in vLLM. January 21, 2026 By By Mathis Felardos Back to.... onlylabs links this event to 1 captured evidence page and 6 related writing signals.","signal_desk":"talking","source_context":{"source_url":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","source_host":"mistral.ai","occurred_at":"2026-01-21T16:00:00+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date","context":null},"context_markers":[{"label":"Lab","value":"Mistral AI","source":"signal"},{"label":"Signal desk","value":"talking","source":"signal"},{"label":"Source host","value":"mistral.ai","source":"source"},{"label":"Notability","value":"Substantive debugging post with low traction","source":"signal"},{"label":"HN","value":"Praise for skilled memory leak diagnosis using SOTA tools.","source":"source"},{"label":"Watch term","value":"Eval methodology","source":"evidence"},{"label":"Watch term","value":"Infrastructure","source":"evidence"}],"evidence_coverage":{"target_pages":1,"captured_pages":1,"readable_pages":1,"capture_methods":["plain"],"missing_page_urls":[],"failed_page_urls":[],"blocked_page_urls":[],"page_urls":["https://mistral.ai/news/debugging-memory-leak-in-vllm/"],"related_signals":6,"has_source_url":true,"latest_page_fetched_at":"2026-06-26T20:23:09.659613+00:00"},"data_business":{"matches":false,"lanes":[],"matched_terms":[],"score":null,"reason":null},"agent_handoff":{"signal_json":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9/signal.json","dossier_json":"https://onlylabs.fyi/labs/mistral/dossier.json","analysis_json":"https://onlylabs.fyi/analysis/mistral/analysis.json","analysis_evidence_json":"https://onlylabs.fyi/analysis/mistral/evidence.json","topic_signals_json":"https://onlylabs.fyi/topics/talking/signals.json","topic_feed":"https://onlylabs.fyi/topics/talking/feed.xml","category_signals_json":"https://onlylabs.fyi/signals.json","data_radar_json":null,"opportunities_json":null},"analysis_playbook":{"objective":"Turn public writing and discussion into a readable map of research themes, product framing, policy posture, launch narratives, and market attention.","evidence_focus":["post title","source URL","captured page text","HN traction","linked model or paper references","publication date"],"extraction_questions":["Which themes are labs choosing to explain publicly?","Which posts are attracting outside discussion?","Which writing reframes a recent release, model, hiring wave, or policy stance?","Which posts mention data, evals, infrastructure, safety, or deployment workflows?"],"signal_questions":["What public theme, launch framing, or research direction does this writing signal expose?","Which themes are labs choosing to explain publicly?","Which posts are attracting outside discussion?","Do the 6 related writing signals show a repeated pattern?"],"output_fields":["org","theme","public_framing","traction","data_business_lane","evidence_url"],"data_business_relevance":"Public writing supplies the narrative layer over raw signals and helps identify which frontier-lab priorities are becoming externally legible.","required_sources":[{"label":"signal_json","url":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9/signal.json","required":true},{"label":"source","url":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","required":true},{"label":"dossier_json","url":"https://onlylabs.fyi/labs/mistral/dossier.json","required":true},{"label":"analysis_evidence_json","url":"https://onlylabs.fyi/analysis/mistral/evidence.json","required":true},{"label":"topic_signals_json","url":"https://onlylabs.fyi/topics/talking/signals.json","required":false},{"label":"data_radar_json","url":null,"required":false}],"expected_output":["one-paragraph source-grounded interpretation","category-specific implication","confidence and missing evidence","recommended next source to inspect"],"prompt_seed":"Using only the linked onlylabs JSON, captured source context, and cited evidence, analyze Mistral AI's writing signal \"Heaps do lie: debugging a memory leak in vLLM.\" for frontier lab strategy."},"semantic_triples":[{"subject":"Mistral AI","predicate":"published","object":"Heaps do lie: debugging a memory leak in vLLM.","text":"Mistral AI published Heaps do lie: debugging a memory leak in vLLM.."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"is classified as","object":"writing signal","text":"Heaps do lie: debugging a memory leak in vLLM. is classified as writing signal."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"belongs to","object":"talking desk","text":"Heaps do lie: debugging a memory leak in vLLM. belongs to talking desk."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has evidence coverage","object":"1 captured evidence page","text":"Heaps do lie: debugging a memory leak in vLLM. has evidence coverage 1 captured evidence page."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has captured page count","object":"1","text":"Heaps do lie: debugging a memory leak in vLLM. has captured page count 1."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has readable page count","object":"1","text":"Heaps do lie: debugging a memory leak in vLLM. has readable page count 1."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has related signal count","object":"6","text":"Heaps do lie: debugging a memory leak in vLLM. has related signal count 6."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has analysis playbook objective","object":"Turn public writing and discussion into a readable map of research themes, product framing, policy posture, launch narratives, and market attention.","text":"Heaps do lie: debugging a memory leak in vLLM. has analysis playbook objective Turn public writing and discussion into a readable map of research themes, product framing, policy posture, launch narratives, and market attention.."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has source host","object":"mistral.ai","text":"Heaps do lie: debugging a memory leak in vLLM. has source host mistral.ai."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has lab","object":"Mistral AI","text":"Heaps do lie: debugging a memory leak in vLLM. has lab Mistral AI."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has signal desk","object":"talking","text":"Heaps do lie: debugging a memory leak in vLLM. has signal desk talking."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has source host","object":"mistral.ai","text":"Heaps do lie: debugging a memory leak in vLLM. has source host mistral.ai."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has notability","object":"Substantive debugging post with low traction","text":"Heaps do lie: debugging a memory leak in vLLM. has notability Substantive debugging post with low traction."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has hn","object":"Praise for skilled memory leak diagnosis using SOTA tools.","text":"Heaps do lie: debugging a memory leak in vLLM. has hn Praise for skilled memory leak diagnosis using SOTA tools.."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has watch term","object":"Eval methodology","text":"Heaps do lie: debugging a memory leak in vLLM. has watch term Eval methodology."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has watch term","object":"Infrastructure","text":"Heaps do lie: debugging a memory leak in vLLM. has watch term Infrastructure."}]},"intelligence":{"signal_desk":"talking","answer":"Mistral AI published Heaps do lie: debugging a memory leak in vLLM.. This talking signal gives public context for research themes, product direction, policy, or launch framing. High-signal details: Substantive debugging post with low traction · Heaps do lie: debugging a memory leak in vLLM. | Mistral AI Engineering Heaps do lie: debugging a memory leak in vLLM. January 21, 2026 By By Mathis Felardos Back to.... onlylabs links this event to 1 captured evidence page and 6 related writing signals.","semantic_triples":[{"subject":"Mistral AI","predicate":"published","object":"Heaps do lie: debugging a memory leak in vLLM.","text":"Mistral AI published Heaps do lie: debugging a memory leak in vLLM.."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"is classified as","object":"writing signal","text":"Heaps do lie: debugging a memory leak in vLLM. is classified as writing signal."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"belongs to","object":"talking desk","text":"Heaps do lie: debugging a memory leak in vLLM. belongs to talking desk."},{"subject":"Heaps do lie: debugging a memory leak in vLLM.","predicate":"has evidence coverage","object":"1 captured evidence page","text":"Heaps do lie: debugging a memory leak in vLLM. has evidence coverage 1 captured evidence page."}]},"signal":{"id":"44261d80-5d0c-425c-b753-7a6f55563bc9","url":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9","json_url":"https://onlylabs.fyi/signals/44261d80-5d0c-425c-b753-7a6f55563bc9/signal.json","source_url":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","title":"Heaps do lie: debugging a memory leak in vLLM.","summary":"Mistral AI published a writing signal. onlylabs watches public writing for research themes, product direction, and model-launch context.","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-01-21T16:00:00+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date","evidence_coverage":{"target_pages":1,"captured_pages":1,"readable_pages":1,"capture_methods":["plain"],"missing_page_urls":[],"failed_page_urls":[],"blocked_page_urls":[],"page_urls":["https://mistral.ai/news/debugging-memory-leak-in-vllm/"]},"facets":{},"traction":{"github_stars":null,"hn_points":7,"hn_comments":1,"hn_story_id":"46707015","hf_downloads":null,"hf_likes":null},"data_radar":null},"primary_evidence_page":{"is_primary":true,"source_match":true,"url":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","final_url":"https://mistral.ai/news/debugging-memory-leak-in-vllm/","title":"Heaps do lie: debugging a memory leak in vLLM.","http_status":200,"content_type":"text/html; charset=UTF-8","capture_method":"plain","fetched_at":"2026-06-26T20:23:09.659613+00:00","bytes":292403,"raw_path":"e93ec6251597dd5437c9962940f63b5260615f629e12b63832b3fd71abe1e8c6.html","content_hash":"3a4ae3dc7edc80cd3f3e2bce209e17eafd7889ceb90b4c6bd5aa3b1dc4cfcf2d","excerpt_chars":1200,"truncated":true,"excerpt":"Heaps do lie: debugging a memory leak in vLLM. | Mistral AI Engineering Heaps do lie: debugging a memory leak in vLLM. January 21, 2026 By By Mathis Felardos Back to Blog 20 min read Share this post Copy to clipboard Copied A few months ago, our team investigated a suspected memory leak in vLLM. At first, we thought the issue would be easy to spot, something confined to the upper layers of the codebase. But the deeper we looked, the more complex it became. This article kicks off our new Engineering Deep Dive series, where we’ll share how we tackle technical investigations and build solutions at Mistral AI. The issue first appeared during pre-production testing of disaggregated serving with one of our frontier models . Memory usage was climbing steadily, but only under specific conditions: with vLLM, with our model Mistral Medium 3.1 , and graph compilation enabled. There were no crashes or errors, just a slow linear increase in system memory of 400 MB per minute on production-like traffic. After a few hours, that would lead to an \"out of memory\" state. What followed was a methodical hunt, starting from high-level Python tools and descending into kernel-level tracing, until we..."},"evidence_pages":[],"related_signals":[{"id":"6659c084-7b80-4fa6-9ddf-35bbee1732c0","url":"https://onlylabs.fyi/signals/6659c084-7b80-4fa6-9ddf-35bbee1732c0","source_url":"https://mistral.ai/news/more-control-over-connectors/","title":"Bringing more control over your connectors","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-06-24T12:00:30+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"},{"id":"7f0b69a8-ccc2-46e9-bbd6-b2d907c58ff5","url":"https://onlylabs.fyi/signals/7f0b69a8-ccc2-46e9-bbd6-b2d907c58ff5","source_url":"https://mistral.ai/news/ocr-4/","title":"Introducing Mistral OCR 4","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-06-23T12:00:48+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"},{"id":"1df2c218-6ea3-4713-8f08-2d94d8dc5a9f","url":"https://onlylabs.fyi/signals/1df2c218-6ea3-4713-8f08-2d94d8dc5a9f","source_url":"https://mistral.ai/news/ai-now-summit-2026/","title":"AI Now Summit 2026","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-05-28T12:00:20+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"},{"id":"cfee09c0-fede-4e0f-9246-4800c0dfdc28","url":"https://onlylabs.fyi/signals/cfee09c0-fede-4e0f-9246-4800c0dfdc28","source_url":"https://mistral.ai/news/vibe-agent/","title":"Vibe gets to work.","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-05-28T12:00:09+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"},{"id":"263b565b-b6bd-4920-bb61-3d24328998b5","url":"https://onlylabs.fyi/signals/263b565b-b6bd-4920-bb61-3d24328998b5","source_url":"https://mistral.ai/news/search-toolkit/","title":"Introducing Search Toolkit","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-05-28T11:42:21+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"},{"id":"4e8a5478-ca7d-4f96-a06c-43d53a3aa21b","url":"https://onlylabs.fyi/signals/4e8a5478-ca7d-4f96-a06c-43d53a3aa21b","source_url":"https://mistral.ai/news/introducing-physics-ai-at-mistral/","title":"Introducing physics AI at Mistral: the foundation for engineering acceleration.","context":null,"kind":{"key":"post_published","label":"Writing"},"org":{"slug":"mistral","name":"Mistral AI","category":"frontier-lab"},"occurred_at":"2026-05-27T12:00:55+00:00","first_seen_at":"2026-06-26T20:22:13.235771+00:00","date_source":"rss.item_date"}]}