←── back to feed
/topics/thinking-machines-lab-inkling-975b-model-release
Thinking Machines Lab Inkling 975B model release
3 items●1 sources●updated 21d ago●trend 0
Thinking Machines Lab released Inkling on July 15, 2026, a 975B-parameter open-weights multimodal mixture-of-experts model with 41B active parameters, 1M-token context window, and native support for text, image, and audio. The model ships under Apache 2.0 and emphasizes controllable thinking effort and customization over raw performance, positioning itself as a foundation for downstream applications rather than a general-purpose frontier model.
- 975B total parameters with 41B active parameters in MoE architecture
- 1M-token context window supporting text, image, and audio inputs
- Apache 2.0 open-weights release on July 15, 2026
- Controllable thinking effort as primary differentiator from competitors
- Company's first model trained from scratch after 18 months of infrastructure development