Release · High impact3 sources
Thinking Machines' Inkling 2 Small claims top spot among western open-weight models
The Decoder · By Matthias Bastian ·

The Brief
- Thinking Machines is releasing Inkling 2 Small, a mixture-of-experts model with 594 billion parameters
- The company claims it is the strongest western open-weight model
- Latent Space reports training took 68 hours on 12T tokens and claims it beats Muse Spark
- The Decoder notes better agentic performance comes at a price, though the source excerpt does not specify the cost
Why it matters
Developers gain a new large open-weight model option from a western lab, but the claimed lead over other open-weight models comes from the company and the sources provide no independent benchmarks.
Who’s affected
- Developers using or evaluating open-weight LLMs
- Teams building agentic AI systems
- #open-weight
- #llm
- #mixture-of-experts
- #thinking-machines
- #agentic-ai
Thinking Machines, the startup from former OpenAI CTO Mira Murati, is releasing Inkling 2 Small, a mixture-of-experts model with 594 billion parameters. The company claims it's the strongest western open-weight model. Better agentic performance comes at a price, though, with…
Timeline
Full coverage · 3 sources
- 4h ago ↗
The Decoder
Thinking Machines' Inkling 2 Small claims top spot among western open-weight models
- 7h ago ↗
Latent Space
[AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark - 17h ago ↗
Hacker News: Front Page
Inkling 2 Small