Thinking Machines Lab is a high-profile AI startup founded by Mira Murati in February 2025. It offers AI solutions and data science services focused on building accessible, customizable, and safe systems. Located in San Francisco, the lab has raised a staggering $2 billion in seed funding and released its first product, Tinker—an API to help developers fine-tune language models more easily.
Origins and Mission
Thinking Machines Lab emerged from broad recognition that powerful AI systems remain opaque and hard to customize. From the start, it aimed to close that gap.
In February 2025, Mira Murati—former OpenAI CTO—launched the lab as a public-benefit corporation, structured for long-term accountability and innovation . Within months, the company raised around $2 billion in seed funding, reaching a valuation near $12 billion, with investors such as Andreessen Horowitz, Nvidia, AMD, Cisco, Jane Street, and even the Albanian government backing it .
The core mission? To make frontier AI systems more understandable, customizable, and generally capable—empowering human–AI collaboration rather than opaque autonomy .
Team & Culture
Thinking Machines Lab assembled a top-tier founding team drawn from elite AI labs:
- OpenAI veterans like John Schulman (Chief Scientist), Barret Zoph (then-CTO), Lilian Weng, Andrew Tulloch, and others joined early .
- Advisory roles included heavyweights such as Bob McGrew and Alec Radford .
- More recent hires include programming Olympiad champion Neal Wu, noted for his high-level coding skills .
That said, the lab has seen turbulence: co-founders Zoph and Luke Metz were fired or departed early in 2026 under messy circumstances; co-founder Andrew Tulloch also left for Meta last year .
Still, leadership remains firmly under Murati, who maintains weighted voting powers and majority control .
Products: Tinker API
The lab’s maiden product, Tinker, debuted October 2025. It’s an API service allowing developers and researchers to fine-tune language models without managing infrastructure.
- Runs jobs on internal clusters, handles scheduling, resource allocation, and failure recovery .
- Uses LoRA to share compute resources across runs and reduce costs .
- Currently in private beta and free to selected users; supports only a handful of open-weight models .
This reflects the lab’s practical-first approach: building tools that amplify research and user experience, grounded in real-world utility.
Values: Open Science, Safety, and Collaboration
Thinking Machines Lab emphasizes transparency and shared progress:
- Committed to publishing code, papers, and blog posts regularly to demystify frontier AI and foster collaborative research .
- Builds multimodal AI systems designed to work with people across text, vision, code, and more .
- Prioritizes AI safety—through best-practices, alignment work, red-teaming, and shared resources—aiming to mitigate misuse and accelerate responsible development .
“We’re making what is otherwise a frontier capability accessible to all, and that is completely game-changing.”
— Mira Murati on the impact of Tinker
Strategic Position and Future Landscape
Thinking Machines Lab sits at a high-stakes intersection in the AI industry:
- The lab is a high-profile challenger to Big Tech—born from OpenAI and built to balance power and public benefit .
- The $2 billion seed makes it one of the most capitalized early-stage companies in history .
- The company’s path forward hinges on delivering actual products and research, while navigating team shakeups and fierce talent competition .
It’s a bold bet on people, infrastructure, and openness. Now, the execution begins.
Conclusion
Thinking Machines Lab is not just another AI startup—it’s a mission-driven lab striving to reshape how frontier AI is built, used, and shared. Through massive funding, elite team formation, and a mismatched mix of ambition and turbulence, it represents both promise and volatility in today’s AI scene.
With its first product, Tinker, the lab shows practical ambition. Now, the real test will be scaling value, weathering talent churn, and delivering safe, transparent AI that lives up to its founding vision.
FAQs
What is Thinking Machines Lab’s main product?
The company’s first offering is Tinker, a managed API that simplifies fine-tuning language models by handling the infrastructure and leveraging techniques like LoRA to lower resource costs.
Who founded Thinking Machines Lab?
Mira Murati, former CTO of OpenAI, launched the company in February 2025 with a mission to democratize AI through customizable, transparent systems.
How did Thinking Machines Lab get funded?
In July 2025, it raised approximately $2 billion in seed funding led by Andreessen Horowitz, with notable backers including Nvidia, AMD, Cisco, Jane Street, and the Albanian government.
Why is AI safety important to the lab?
The lab centers AI safety as a core tenet—through open practices, red-teaming, and publishing guidelines—to ensure systems are aligned and reduce misuse risk.
Has there been leadership turnover?
Yes. Co-founders Barret Zoph and Luke Metz left or were fired early in 2026 amid internal conflicts, while Andrew Tulloch departed for Meta in October 2025.
Leave a comment