Inkling 2 Mini solves problems with fewer words, outperforms on coding tests
thinking machines releases inkling 2 mini, a looped transformer model with 26.4b parameters, outperforming open models under 100b on agentic coding tests, available under apache 2.0.
thinking machines released inkling 2 mini, a looped transformer model with 26.4 billion parameters, of which 5.4 billion are active per token. it runs 12 of its 24 layers three times in a row, allowing for internal reasoning without generating more text. the model processes text and images, has a 1-million-token context window, and is licensed under apache 2.0. inkling 2 mini leads open models under 100 billion parameters on agentic coding tests, scoring 48.2 percent on deepswe v1.1 and 46.5 percent on vals index v2.2. it also scores 20 on artificial analysis's intelligence index, tying with qwen3.8 27b at lower cost and higher speed. the model can run locally on macs with at least 32 gb of memory and on nvidia's dgx spark.