{"type":"post","id":"ai-r-d-relevant-model-capabilities-are-improving-3b98d6","slug":"ai-r-d-relevant-model-capabilities-are-improving-3b98d6","title":"AI R&D-relevant model capabilities are improving rapidly and are getting harder to measure","description":"METR’s November 2026 update estimates that measured AI task horizons across its combined Time Horizon suites are doubling about every 3.6 months.","retrieval_nugget":"METR’s November 2026 update estimates that measured AI task horizons across its combined Time Horizon suites are doubling about every 3.6 months. Its broader public-model index does not show a strong overall acceleration, while judgment-heavy work remains a notable weakness. For builders, the result separates fast progress on testable work from reliable autonomy on work that is difficult to verify.","published_at":"2026-11-21","updated_at":"2026-11-21","record_date":"2026-11-21","date_kind":"published_at","topics":["ai"],"entities":["METR","Time Horizon 1.1","TH2","MirrorCode","GPT-6 Astra","Opus 5.5"],"related_patterns":["verification-bandwidth-is-the-scarce-resource"],"novelty_label":"new_fact","what_changed":null,"prior_refs":[],"source_url":"https://metr.org/blog/2026-11-20-q4-2026-capability-update","source_urls":["https://metr.org/blog/2026-11-20-q4-2026-capability-update"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:ai-r-d-relevant-model-capabilities-are-improving-3b98d6","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/ai-r-d-relevant-model-capabilities-are-improving-3b98d6/","markdown":"https://newruntime.com/posts/ai-r-d-relevant-model-capabilities-are-improving-3b98d6.md","json":"https://newruntime.com/posts/ai-r-d-relevant-model-capabilities-are-improving-3b98d6.json"},"markdown_url":"https://newruntime.com/posts/ai-r-d-relevant-model-capabilities-are-improving-3b98d6.md","next_reads":[{"type":"topic","path":"/topics/ai/","url":"https://newruntime.com/topics/ai/","title":"Ai","reason":"Explore the ai topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/ai.md","json_url":"https://newruntime.com/topics/ai.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}
