Thinking Machines Unveils 594B-parameter Inkling 2 Small AI Model
Article excerpt
Highlighted: the sentence this signal was extracted from
Thinking Machines released Inkling 2 Small, a 594-billion-parameter mixture-of-experts model with 17 billion active parameters, image input support and variable reasoning effort. The company said the model initially will be available through its Tinker platform with closed weights. It plans to release the full weights under an Apache 2.0 license next week after external safety testing and red-teaming are completed. Inkling 2 Small is available in Tinker Playground, and developers can fine-tune and sample the model on Tinker. Checkpoints will be downloadable when the weights are released. Don't miss out on the latest insights, trends, and analysis in the world of data, technology, and startups. Subscribe to our newsletter and get exclusive content delivered straight to your inbox. Thinking Machines said Inkling 2 Small scored 31 on the Artificial Analysis Intelligence Index, exceeding its earlier models. The company attributed the improvement to more efficient pre-training and a large unified reinforcement-learning run. The model was trained on hundreds of thousands of long-horizon coding tasks across multiple agent harnesses and context-compaction approaches. Thinking Machines said it used agentic reinforcement-learning graders that assessed how a model approached a programming problem as well as whether its code passed tests. On FrontierCode v2.0, Inkling 2 Small scored...
Keep reading with a free account
The rest of this article, and every signal for Thinking Machines Lab, is in your free account.
