Mira Murati spent years helping build some of the most powerful closed AI systems on the planet at OpenAI. Now she’s giving it all away for free.
Thinking Machines Lab, the AI startup Murati founded after departing OpenAI, released its first in-house model called Inkling on July 15. The model packs 975 billion parameters, processes text, images, audio, and video simultaneously, and ships under the permissive Apache 2.0 license. Anyone can download the weights from Hugging Face and do essentially whatever they want with them.
What Inkling actually is
Inkling uses a Mixture-of-Experts architecture, a design pattern where only a subset of the model’s total parameters activate for any given task. While the full model weighs in at 975 billion parameters, only 41 billion are active at any one time. The training run consumed 45 trillion tokens and was built from scratch on NVIDIA GB300 NVL72 hardware. The model supports a context window of 1 million tokens.
Thinking Machines Lab is also releasing a platform called Tinker alongside Inkling, designed to let researchers and developers customize the model for specific use cases.
Performance benchmarks, including SWE-bench Verified, suggest Inkling is competitive but not the outright leader in every category.
The business behind the model
Thinking Machines Lab was founded in February 2025. In that time, the company raised $2 billion in funding at a $12 billion valuation. Reports suggest ongoing discussions could push that figure toward $50 billion.
Key staff departures in January 2026 briefly raised questions about the lab’s stability, but the Inkling release appears to answer those concerns definitively. Shipping a model of this scale on a timeline of roughly 18 months from founding is an aggressive pace by any standard.
Why crypto investors should pay attention
Inkling itself has no token, no blockchain integration, and no crypto component. Murati’s lab is a pure AI play.
A 975 billion parameter open model with multimodal capabilities expands the design space for what decentralized AI applications can actually do. Decentralized compute marketplaces that aggregate GPU resources for inference, projects like Akash, Render, and io.net, stand to benefit from growing demand to run models of this scale outside of centralized cloud providers.
The valuation trajectory of Thinking Machines Lab also offers a useful benchmark for crypto-AI projects. If a centralized open-source AI company can command a $12 billion valuation (potentially heading to $50 billion), the market is clearly pricing in massive value for open AI infrastructure.
The competitive dynamic to watch is whether Inkling’s Apache 2.0 license creates a wave of derivative models, fine-tuned versions optimized for specific verticals like trading, DeFi risk analysis, or smart contract auditing.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
