{"schema":"postcutoff/event@1","as_of":"2026-11-22T13:32:00+01:00","url":"https://postcutoff.com/e/2026-11-20-thinking-machines-inkling-2-mini-looped/","md":"https://postcutoff.com/e/2026-11-20-thinking-machines-inkling-2-mini-looped/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-11-20-thinking-machines-inkling-2-mini-looped","date":"2026-11-20","date_precision":"day","short_title":"Inkling 2 Mini: open 26B looped model","deck":null,"takeaway":"On 20 Nov 2026 Thinking Machines Lab released Inkling 2 Mini, a 26.4B-parameter (3.4B active) mixture-of-experts \"looped\" transformer that runs its middle layers three times per token and fits in about 20 GB, with Apache-2.0 weights on Hugging Face.","category":"model-release","category_label":"Model releases","importance":3,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":[{"n":1,"title":"Thinking Machines: Inkling 2 Mini","url":"https://thinkingmachines.ai/news/inkling-2-mini","type":"official","group":"primary","domain":"thinkingmachines.ai"},{"n":2,"title":"Hugging Face: thinkingmachines/Inkling-2-Mini","url":"https://huggingface.co/thinkingmachines/Inkling-2-Mini","type":"code","group":"primary","domain":"huggingface.co"},{"n":3,"title":"Hugging Face blog: Welcome Inkling 2 Mini","url":"https://huggingface.co/blog/inkling2mini","type":"docs","group":"primary","domain":"huggingface.co"},{"n":4,"title":"Mira Murati on X","url":"https://x.com/miramurati/status/2123872534970208421","type":"official","group":"primary","domain":"x.com"},{"n":5,"title":"ARC Prize on X: Inkling 2 Mini on ARC-AGI (Verified)","url":"https://x.com/arcprize/status/2123885049091440920","type":"discussion","group":"reactions","domain":"x.com"}],"official":4,"filed":"2026-11-21","updated":"2026-11-22","orgs":["Thinking Machines Lab"],"title":"Thinking Machines releases Inkling 2 Mini, an open-weights 26B looped transformer that runs on a laptop","summary":"On 20 Nov 2026 Thinking Machines Lab released Inkling 2 Mini, a 26.4B-parameter (3.4B active) mixture-of-experts \"looped\" transformer that runs its middle layers three times per token and fits in about 20 GB, with Apache-2.0 weights on Hugging Face. It leads open models under 100B parameters on DeepSWE and the Vals Index. Mira Murati said its reasoning stays monitorable at this scale but \"this may not stay the case\"; the lab also released a variable-loop research model for study.","key_facts":["Size: 26.4B total, 3.4B active per token; 24 stored layers in four six-layer blocks, the two middle blocks run three times (48 layer applications per token); mostly NVFP4 weights, ~20 GB","Benchmarks (lab): DeepSWE v1.1 48.2%, Vals Index v2.2 46.5% (best open model under 100B), AA Index v5.0 20.1, SWE-bench Verified 79.2%, BrowseComp 79.0%; trails Qwen3.8-27B on ComplexConstraints (33.4%)","Artificial Analysis (Nov 20, via Techmeme): matches Qwen3.8 27B (xhigh) on its Intelligence Index with about 5x fewer active parameters","Availability: open weights, Apache-2.0 (huggingface.co/thinkingmachines/Inkling-2-Mini); Tinker; SGLang, vLLM, Ollama, llama.cpp (GGUF)","Research model TML-d24-varloop-research (1–5 passes): no-CoT time horizon rose from 16 s at one pass to 37 s at three","Mira Murati on X (506k views): 'At this scale, its thought process remains monitorable; this may not stay the case'; the lab reports no meaningful loss of reasoning monitorability","ARC Prize verified scores (20 Nov): ARC-AGI-2 49.3% at $0.14 per task, ARC-AGI-1 84.5% at $0.06; 47% cheaper per task than Inkling 2 Small on ARC-AGI-2 but 32 points lower"],"key_numbers":[],"tags":["open-weights","looped-transformer","moe","on-device","thinking-machines","monitorability"],"science":null,"body_md":"## What happened\n\nTwo weeks after Inkling 2 Small (594B), Thinking Machines released a small sibling for local use. Instead of more layers it reuses the middle of the network: each token passes through the two central blocks three times, which the lab says lets the model \"think more and write less\".\n\n## Why it matters\n\nLooping adds hidden computation that does not appear as visible chain-of-thought. The lab says reasoning is still monitorable at this size, and released a variable-loop model so others can study the point at which that stops being true.","disputed":[],"related":[{"id":"2026-11-05-thinking-machines-inkling-2-small","url":"https://postcutoff.com/e/2026-11-05-thinking-machines-inkling-2-small/","date":"2026-11-05","date_precision":"day","short_title":"Thinking Machines Lab releases Inkling 2 Small on Tinker","deck":"Apache-2.0 weights followed on Nov 12 after safety testing","takeaway":"With the weights out (Nov 12) it is the strongest US open-weights model by Artificial Analysis' estimate; our open-weights pick stays Kimi K3.1 Pro, which scores higher (AA v5 46 vs ~31).","category":"model-release","category_label":"Model releases","importance":4,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":7,"official":3,"filed":"2026-11-06","updated":"2026-11-15","orgs":["Thinking Machines Lab"]}],"people":[{"id":"mira-murati","name":"Mira Murati","url":"https://postcutoff.com/person/mira-murati/"}],"posts":[{"id":"x-miramurati-2123872534970208421","title":"“Inkling 2 Mini is a looped transformer that runs on a laptop. Looping lets a model think…”","url":"https://postcutoff.com/p/x-miramurati-2123872534970208421/"}],"videos":[],"models":[{"id":"inkling-2-mini","name":"Inkling 2 Mini","url":"https://postcutoff.com/m/inkling-2-mini/"}],"changes":[{"date":"2026-11-22","type":"edit","text":"ARC Prize verified scores (news agent)"},{"date":"2026-11-21","type":"filed","text":"Created (news agent)"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-11-21","run":null,"sources_read":null,"updated":"2026-11-22","human_review":null,"version":null},"gaps":[{"model_id":"gpt-6-2-astra","name":"GPT-6.2 Astra","cutoff":"2026-05","days_after":173,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":143,"in_training_data":false},{"model_id":"gemini-4-hydrogen","name":"Gemini 4 Hydrogen","cutoff":"2026-03","days_after":234,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":173,"in_training_data":false}],"short_url":null}