

Inkling 2 Mini
#8 in Lütte & Edge-ModelleThinking Machines · v2 · mini · siet 20.11.2026 · 5× · tolest 22. Nov. 2026
49
Momentum
Inkling 2 Mini is an open-weights, compact Mixture-of-Experts model from Thinking Machines Lab with 26.4B total and 3.4B active parameters. It uses a looped transformer architecture in which the middle layers run three times per token and is designed for local use on laptops. The model is released under Apache 2.0 on Hugging Face and is also available via Tinker.
Momentum-Verloop
24.08.22.11.
Features
| Key Benchmark (%) | 48.2% DeepSWE v1.1; 77.9% SWE-Bench Pro v2 (Hard) (per Unsloth) |
| Context Window (Tokens) | Up to 1M tokens |
| License | Apache 2.0 |
| Multimodality | Input: text and images; output: text |
| Platform | Hugging Face (open weights), Tinker (inference/fine-tuning), local e.g. via Ollama (Mac with 32 GB RAM) or GGUF |
| Price per 1M Tokens | Input $0.15 / Output $0.50 (cache hits $0.015) via Tinker |
| Release Date | November 20, 2026 |