Summary
According to a post on X by @ollama, Thinking Machines' looped 26B model, Inkling 2 Mini, is available on Ollama 0.40.17 (pre-release) and can run locally on Macs with 32GB or more of memory. The model is also live in LMArena's Agent, WebDev and Text arenas, with scores still to come. Unsloth has released quants that the post says keep over 90% top-1% accuracy.
AI-assisted summary based on the listed source.
What happened
Thinking Machines' looped 26B model is now on Ollama 0.40.17 (pre-release) and runs locally on Macs with 32GB+ memory, is live in LMArena's Agent, WebDev and Text arenas with scores coming, and has Unsloth quants that keep over 90% top-1% accuracy. Why it matters: Fast availability in popular local and evaluation...
Why it matters
Developers who want to run an open-weight model locally can now try this one through Ollama on Macs with 32GB or more of memory. Evaluators can follow its LMArena results once scores are published. The post frames the fast rollout across local and evaluation tools as a quick route to real developer use for a US open-weight model, so the next step is to watch the arena scores and how the pre-release Ollama build performs.
Signal Intelligence
Signal Strength 90%
Technical label WATCH
Public Interest 54
Category OPEN SOURCE
Reader Depth TECHNICAL
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
Public Interest components
Recognizable Entity Score 73
Practical Impact Score 28
Novelty Interest Score 94
Consequence Score 18
Curiosity Score 16
Shareability Score 66
Why this is here
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).