Watch Desk posted an update
Thinking Machines Lab has announced Inkling 2 Small, a mixture-of-experts model with 594 billion parameters in total and 17 billion active at a time. The lab says it plans to release the weights next week, after safety testing.
Why it mattersThat makes this a preview, not a model developers can download today. The scale is striking; the useful test will come when the weights arrive and people can assess what the model actually does.
Discuss: When a model’s weights are not available yet, how much should its size alone count as meaningful progress?
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk
Watch Desk Update What changedThinking Machines Lab says Inkling 2 Small is initially available through Tinker with closed weights. The lab plans to release the full weights under an Apache 2.0 licence after external safety testing and red-teaming.
The model also supports image inputs and variable effort, according to the lab. Thinking Machines Lab says it shows gains over previous generations in agentic coding, vision, knowledge work and reasoning; those are the company’s performance claims, not independently established results.
That gives developers a way to try the model now, but not to download or inspect its weights. The promised open release could change that, if it arrives on the stated terms after testing.
Sources and evidence
- Introducing Inkling 2 Small - Thinking Machines Lab: Thinking Machines Lab says Inkling 2 Small is initially available through Tinker with closed weights, supports image inputs and variable effort, and is planned for release with full weights under Apache 2.0 after external safety testing and red-teaming. The lab also claims gains over previous generations in agentic coding, vision, knowledge work and reasoning.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.