Thinking Machines Releases 975B-Parameter Open-Weights Multimodal Model
Key Takeaways
- ▸Thinking Machines releases 975B-parameter open-weights model with multimodal capabilities (text, image, audio)
- ▸Model features efficient architecture with only 41B active parameters, optimizing computational requirements
- ▸Full weights available on Hugging Face for community fine-tuning and experimentation
Summary
Thinking Machines, led by former OpenAI CTO Mira Murati, has released a new open-weights large language model featuring 975 billion total parameters with 41 billion active parameters. The model is available on Hugging Face and supports text, image, and audio processing, enabling community fine-tuning and experimentation across multiple modalities.
The open-weights release democratizes access to advanced AI capabilities, allowing researchers, developers, and organizations to adapt the system for their specific use cases without proprietary constraints. The efficient architecture—utilizing only 41 billion active parameters despite the 975 billion total parameter count—reflects careful model design optimized for both capability and computational efficiency.
This release positions Thinking Machines as a significant player in the open-source AI ecosystem following Murati's departure from OpenAI. By providing full access to model weights across text, image, and audio modalities, the company signals its commitment to advancing AI innovation through openness while maintaining high performance standards.
- Positions Thinking Machines as a major contributor to the open-source AI ecosystem
