Thinking Machines amps up its bet against one-size-fits-all AI with its first open model, Inkling

What happened

Thinking Machines Lab released Inkling on July 15, 2026 — the startup's first open-weights model. Founded by former OpenAI CTO Mira Murati, the company is valued at $12B and backed by Nvidia.

Context and impact

Inkling is a direct challenge to the dominance of closed models from OpenAI, Anthropic, and Google. Thinking Machines bets that enterprises want AI they can customize, not just rent. Weights are freely available on Hugging Face and the model can be fine-tuned via their Tinker platform. Inkling is not the strongest model on benchmarks but targets better performance-per-cost in specific use cases through customization.

Details

• Architecture: Mixture-of-Experts, 975B total parameters, ~41B active per inference
• Training data: 45 trillion tokens of text, images, audio, and video
• Multimodality: native support across all four input modalities
• Fine-tuning: available through Thinking Machines' Tinker platform
• Inkling-Small: lighter model in preparation — weights to be released after testing
• Primary source: thinkingmachines.ai/news/introducing-inkling/