Thinking Machines makes Inkling 2 Small available on Tinker
Inkling 2 Small uses a 594B-total, 17B-active MoE architecture with image input and long-horizon agent capabilities; it is available today on Tinker with closed weights.
AT A GLANCE
Key points
- The MoE model has 594B total and 17B active parameters.
- Inkling 2 Small supports image input and long-horizon agent tasks.
- Tinker access is live, but the model weights were not released at launch.
- The company targets an Apache 2.0 weight release after safety testing.
What is changing?
Thinking Machines Lab made its second-generation Inkling 2 Small model available through Tinker on November 5. The Mixture-of-Experts model has 594 billion total parameters and 17 billion active parameters, and supports image inputs and variable effort. The company reports gains over its first generation, particularly in long-horizon autonomous work, agentic coding, vision tasks, and knowledge work.
The model can be tried today in Tinker Playground, while Tinker also supports fine-tuning and sampling. Its weights were still closed at launch. Thinking Machines says it intends to release the full weights under Apache 2.0 next week after external safety testing and red-teaming are complete. Open-weight access therefore should not be treated as available yet; teams should evaluate current Tinker access separately from the planned weight release.
Explore the original sources
Source date: 5 November 2026.
Published by AI Radar: 6 Nov 2026, 04:04 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…