Inkling 2 Mini architecture: 24 layers, with the middle 12 looped three times
thinkymachines · x · 2026-11-21
Thinking Machines detailed Inkling 2 Mini's architecture: 24 layers total, with the middle 12 looping three times per token and the first and last six running once. Looping lets the model reason more while writing less, which has raised monitorability concerns the team studied as part of the release.
More from Models
- Inkling 2 Mini runs on 32GB Macs via Ollama and on NVIDIA DGX Spark — thinkymachines · 2026-11-21
- Thinking Machines releases Inkling 2 Mini, a 26B looped transformer that runs on laptops — thinkymachines · 2026-11-21
- Thinking Machines: more loop passes boost no-CoT capability without hurting monitorability — thinkymachines · 2026-11-21
- OpenAI emails API users 50% off new frontier model 6.2-Astra for 8 weeks — soumitrashukla9 · 2026-11-21
- Thinking Machines releases Inkling 2 Mini, a looped transformer topping sub-100B open-weights coding benchmarks — SonglinYang4 · 2026-11-21
- OnlyModels Model Fight 17: Qwen distills dominate, only 5/72 APEX claims held — Dave_Maynor · 2026-11-21