Inkling
Inkling is an open-weights multimodal Mixture-of-Experts language model released by Thinking Machines Lab with 975 billion total parameters and 41 billion active parameters per token, trained on 45 trillion tokens and licensed under Apache 2.0. The model supports a 1 million token context window across text, images, and audio, with a smaller 276 billion parameter variant also available, and is positioned for enterprise customization and fine-tuning through the Tinker platform rather than as a frontier model.
Updated 3 August 2026
Specifications
No specifications recorded yet.
Latest developments
Thinking Machines Lab Releases Inkling-Small: A 276B Total, 12B Active Open Weights Multimodal MoE Model
MarkTechPost · 18 hours ago ·
13
Latest open artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 show the utility of open models on the Pareto frontier
Interconnects · 1 day ago ·
27
The Sequence AI of the Week #899: Inside Inkling: A Trillion-Parameter Model That Only Wakes Up 41 Billion at a Time
TheSequence · 1 week ago ·
3
The Sequence Radar #897: Last Week in AI: China, Compression and the Open-Model Race
TheSequence · 2 weeks ago ·
36
Inkling: Our open-weights model
Simon Willison · 2 weeks ago ·
39
Deep Learning Weekly: Issue 464
Deep Learning Weekly · 2 weeks ago ·
10
I've got an Inkling
Ben's Bites · 2 weeks ago ·
23
Inkling: Our Open-Weights Model
TLDR Dev · 2 weeks ago ·
15
Q3 2026
Multiple AI companies release competitive open-weight models as industry trend shifts toward openness Open source release
Moonshot AI releases Kimi K3, a 2.8 trillion parameter open-weight model achieving frontier-class performance Open source release
Thinking Machines Lab releases Inkling, a 975B-parameter open-weights multimodal model under Apache 2.0 license Open source release
Thinking Machines Lab releases Inkling, a 975B-parameter open-weights multimodal mixture-of-experts model Open source release
Thinking Machines Lab releases Inkling, a 975-billion-parameter open-weight AI model Open source release
Thinking Machines releases Inkling, an open-source multimodal language model with 975 billion parameters Open source release
- Thinking Machines Lab Releases Inkling-Small: A 276B Total, 12B Active Open Weights Multimodal MoE Model
- Latest open artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 show the utility of open models on the Pareto frontier
- The Sequence AI of the Week #899: Inside Inkling: A Trillion-Parameter Model That Only Wakes Up 41 Billion at a Time
- The Sequence Radar #897: Last Week in AI: China, Compression and the Open-Model Race
- Inkling: Our open-weights model
- Deep Learning Weekly: Issue 464
- I've got an Inkling
- Inkling: Our Open-Weights Model
- Mira Murati's AI Startup Releases First Model in Bid to Loosen AI Giants' Grip
- Thinking Machines released Inkling, an open-weight customizable multimodal model
- [AINews] Thinky's Inkling: 975B-A41B multimodal, new best American Apache 2.0 open model (with Inkling-Small, 276B-A12B)
- Thinking Machines Lab Releases Inkling: A 975B-Parameter Open-Weights Multimodal MoE With 41B Active Parameters And Controllable Thinking Effort
- Thinking Machines Lab Drops Its First Model
- Thinking Machines amps up its bet against one-size-fits-all AI with its first open model, Inkling
- Together AI brings Thinking Machines Lab’s new model Inkling on day 0
- Welcome Inkling by Thinking Machines
Relationships
Products & technology
- Thinking Machines Lab develops this model · 8 sources
- Thinking Machines develops this model · 5 sources
- Integrated with Tinker · 1 source
- Integrated with Hugging Face · 1 source
- Thinking Machine develops this model · 1 source
- Inkling-Small derived from this model · 1 source
- Together AI deploys this model · 1 source