Thinking Machines releases Inkling 2 Small, with open weights planned for next week

Thinking Machines Lab introduced Inkling 2 Small on its website, making the model available through Tinker and Tinker Playground. The Mixture of Experts model has 594 billion total parameters, activates 17 billion parameters per request, accepts image inputs and supports variable effort levels.

Inkling 2 Small scored 31 on the Artificial Analysis Intelligence Index, compared with 13 for Inkling Small. The company reports gains in coding, visual reasoning and knowledge work, including scores of 86.3% on SWEBench Pro v2 Hard and 1527 Elo on AA Briefcase.

Artificial Analysis said Inkling 2 Small will be the highest scoring open weights model outside China once its weights are released. The model uses about 150,000 output tokens per index task, roughly four times as many as Inkling Small, and costs $1.18 per task. Among open models, only DeepSeek V4.1 Flash and Z.ai's GLM-5.3 score higher, at 32 points each, Trending Topics reported.

The model currently has closed weights. Thinking Machines Lab plans to release the full weights under the Apache 2.0 license next week after external safety testing and red teaming. Inkling 2 Small is also available for fine tuning and sampling on Tinker, with downloadable checkpoints planned alongside the weights release.