Mira Murati's Thinking Machines Lab Debuts Open-Weight AI Model Inkling with 975 Billion Parameters
By
Mr Bagel
Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, released its first open-weight model called Inkling on Wednesday, according to TechCrunch and Reuters via KELO-AM. The release marks the company's first major public milestone after 18 months of building AI infrastructure largely out of public view, TechCrunch reported.
Inkling uses a mixture-of-experts architecture with 975 billion total parameters but activates only about 41 billion per task, making it efficient, TechCrunch detailed. The model was trained on 45 trillion tokens and supports multimodal capabilities including video and audio, spidits.com noted, while developersdigest.tech added that it offers a 1 million token context window.
"The open-weight approach contrasts with the closed models from major competitors like OpenAI, Anthropic, and Google." :: TechCrunch This strategic choice positions Thinking Machines Lab in the growing camp of AI companies that prioritize accessibility and transparency over proprietary secrecy.
"A new American open-weights frontier model with multimodal capabilities, 1M token context, and competitive benchmarks." :: developersdigest.tech Early community reactions on Hacker News, as covered by developersdigest.tech, have focused on the model's competitive benchmark performance and its potential to democratize advanced AI research.
Together AI announced it would host Inkling on its platform on day zero, according to Together, signaling rapid infrastructure support for the new model. The launch represents a significant step for Murati's startup, which has been working to challenge the dominance of major players with a more open approach to AI development.
The reporting
32 outlets covered this story. Each links to the original.

Baker's Take
Comments
Sign in to join the conversation.
No comments yet. Be the first.