# Thinking Machines Lab launched Inkling 2 Small, a 594B-parameter MoE (17B active) you can try on Tinker today, with Apache 2.0 weights planned next week after safety testing.

> Thinking Machines Lab launched Inkling 2 Small, a 594B-parameter MoE on Tinker today, with Apache 2.0 weights planned next week after safety testing.

Published 2026-11-05T17:18:03-08:00 · News · Canonical: https://news.littlewolflabs.com/posts/2026-11-05-inkling-2-small

**Thinking Machines Lab launched Inkling 2 Small, a 594B-parameter MoE (17B active) you can try on Tinker today, with Apache 2.0 weights planned next week after safety testing.**

"Today we are introducing our second model generation with Inkling 2 Small," Thinking Machines wrote on Thursday. "It is a Mixture-of-Experts model with 594B total parameters, 17B active, and supports image inputs and variable effort" ([Thinking Machines](https://thinkingmachines.ai/news/inkling-2-small/)). The lab says the model "far surpasses our previous models’ scores on the Artificial Analysis Intelligence Index with a score of 31," with gains in agentic coding, vision, knowledge work and reasoning. In the lab’s own comparison table, Inkling 2 Small’s AA Index (v5.0) is 30.6, nearly matching Gemini 3.8 Flash at 30.8. Artificial Analysis posted that the model scores 31 on its Intelligence Index and that releasing the weights "would make it the highest-scoring open weights model developed outside China based on current rankings" ([Artificial Analysis on X](https://x.com/ArtificialAnlys/status/2118493376186388530)). Weights start closed: the model "will initially be available through Tinker, with closed weights," and the lab intends "to make the full weights available under an Apache 2.0 license next week after the completion of external safety testing and red-teaming." You can "try Inkling 2 Small on Tinker Playground today," and it is also up for fine-tuning and sampling on Tinker. Tinker’s pricing table lists Inkling 2 Small at $0.30 per million prefill tokens ($0.03 cached) and $1.00 per million sample tokens for the 64K and 256K serverless IDs ([Tinker pricing](https://tinker-docs.thinkingmachines.ai/tinker/models/models_and_pricing/)). Demos on the launch page include iterative website and game building, a Blender paper-and-origami scene, and long-horizon autoresearch on Tinker.

*Try it:* if you fine-tune on Tinker or want a soon-to-be Apache open-weights option, sample Inkling 2 Small in the Playground and watch for the weight drop next week after the safety pass. Treat the AA "31" claim as the lab’s rounded figure; their table shows 30.6 next to Gemini 3.8 Flash’s 30.8.