Thinking Machines Lab

Thinking Machines Lab Inc. is an AI startup founded by Mira Murati in February 2025. Based in San Francisco, the company operates as a public benefit corporation, aiming to make AI systems more accessible, customisable, and powerful for diverse users. Its founding team includes experienced researchers and engineers from top labs, such as OpenAI, Meta AI, and Mistral AI. Among the founding members are Barret Zoph, Lilian Weng, and John Schulman, all from OpenAI.

In July 2025, Thinking Machines Lab secured $2 billion in funding, with a valuation of $12 billion. The funding round was led by Andreessen Horowitz, with investors including Nvidia, AMD, Cisco, Jane Street, and the Albanian government. This financial support has been essential for the company's research and development.

In October 2025, the company launched Tinker, an API that allows users to fine-tune open-weight models. Tinker is part of the company's mission to offer more flexible and customisable AI tools to various users, from developers to entrepreneurs.

Thinking Machines Lab focuses on human-AI collaboration rather than solely developing autonomous systems. It aims to build multimodal AI that can adapt to human expertise and support a broader range of applications. The company also prioritises building reliable, secure infrastructure for long-term productivity and safety. It is committed to advancing AI safety by sharing best practices and engaging in real-world testing.

Thinking Machines Lab Inc. is an American artificial intelligence startup founded in February 2025 by Mira Murati, the former chief technology officer of OpenAI. The company, headquartered in San Francisco, operates as a public benefit corporation. Its mission is to bridge the gaps in AI research and development, particularly focusing on creating more accessible, customisable, and powerful AI systems.

Upon its founding, Thinking Machines Lab quickly attracted attention for its ambitious vision and its leadership team. The company hired researchers and engineers from top AI labs, including OpenAI, Meta AI, and Mistral AI. Among its founding team members were Barret Zoph, former VP of Research at OpenAI, Lilian Weng, former VP of OpenAI, and John Schulman, a co-founder of OpenAI. The company’s advisory board included key figures such as Bob McGrew and Alec Radford, both of whom had significant roles at OpenAI.

In July 2025, Thinking Machines Lab completed an early-stage funding round, raising $2 billion at a valuation of $12 billion. The round was led by Andreessen Horowitz and included investors such as Nvidia, AMD, Cisco, and Jane Street. The Albanian government, Mira Murati's country of origin, also participated, investing $10 million in the company. This funding has been crucial for scaling the company’s research and development efforts.

On October 1, 2025, Thinking Machines Lab introduced Tinker, an API designed for fine-tuning language models. This product enables users to submit jobs for fine-tuning open-weight models, which the company processes using its internal infrastructure. Tinker is a part of the company's broader strategy to provide accessible and customisable AI tools to a wide range of users, including AI developers, entrepreneurs, and business analysts.

Thinking Machines Lab has set itself apart from other AI companies by emphasising human-AI collaboration rather than focusing solely on fully autonomous systems. The company’s approach is centred on building multimodal AI systems that can work with humans in diverse fields, making AI more adaptable, flexible, and personalised. The company also places a strong emphasis on research productivity, reliability, and security, building its infrastructure to support long-term goals.

The company is contributing to AI safety and promoting best practices for building safe AI systems. Its approach to safety combines proactive research with real-world testing, aiming to prevent the misuse of its models while allowing users the freedom to innovate. As part of this commitment, Thinking Machines Lab plans to share its research and models with the broader AI community to advance the industry’s understanding of AI safety.

Thinking Machines Lab's mission is to make artificial intelligence (AI) more accessible, customisable, and powerful. The company aims to bridge the gaps in AI understanding and usage by creating AI systems that work collaboratively with humans, rather than being fully autonomous. It focuses on providing flexible, adaptable, and personalised AI solutions that can be applied across various fields. The company is committed to advancing AI technology, ensuring its products are reliable, secure, and available for everyone to use, regardless of their technical expertise.

Thinking Machines Lab envisions a future where AI is integrated into every aspect of human life. The company strives to build AI systems that can understand and support human expertise in a wide range of applications. By developing multimodal AI, the company aims to improve communication, capture more information, and help solve real-world challenges. Thinking Machines Lab also aims to advance AI safety and ethics, ensuring that AI systems are built responsibly and can be used safely. The company’s goal is to make AI work for everyone, enhancing productivity, creativity, and innovation across industries.

Thinking Machines Lab has rapidly gained recognition in the AI industry for its innovative approach to AI development. The company’s founding team, which includes former OpenAI experts, has made significant contributions to the field of artificial intelligence. It has been praised for its commitment to making AI more accessible and customisable. Additionally, the company’s successful funding rounds, including a $2 billion investment led by Andreessen Horowitz, have reinforced its standing in the AI sector. Thinking Machines Lab’s products, like Tinker, have been recognised for their potential to advance AI’s practical applications across various industries.

Thinking Machines Lab offers a range of innovative products and services designed to make artificial intelligence (AI) more accessible, customisable, and effective for users across various sectors. 

Tinker – Fine-Tuning API: One of the key offerings from Thinking Machines Lab is Tinker, an API for fine-tuning language models. This service allows users to customise open-weight models to better suit their specific needs. With Tinker, users can submit jobs for fine-tuning and run them on the company’s internal clusters, leveraging its powerful infrastructure. This API is particularly valuable for developers, businesses, and researchers who require more tailored AI solutions without needing to set up complex systems on their own. By making fine-tuning more accessible, Tinker helps users adapt AI to different applications, from business insights to personalised content creation.

Customisable AI Models: Thinking Machines Lab focuses on building AI models that can be customised to meet the needs of diverse industries. Unlike traditional AI systems, which often operate in a one-size-fits-all manner, the company prioritises creating adaptable solutions. Their models are designed to integrate smoothly into various workflows, enabling users in fields like science, engineering, and business to create AI systems that understand and complement their unique goals. Whether for data analysis, language processing, or complex scientific research, these customisable models help companies develop AI-driven solutions tailored to their specific challenges.

Multimodal AI Systems: The company is heavily invested in building multimodal AI systems. These systems are designed to process and integrate data from different sources and formats, such as text, images, and video, to improve decision-making and communication. Multimodal AI enables more natural, human-like interactions with AI, supporting deeper integration into real-world environments. By expanding the capability of AI to handle diverse types of input, Thinking Machines Lab seeks to enhance the efficiency of tasks that require a combination of sensory data, making the technology more intuitive and versatile.

Research and Development Collaboration: In addition to its product offerings, Thinking Machines Lab provides a platform for collaboration in AI research and development. The company believes that scientific progress is most effective when shared, which is why they frequently publish technical papers, blog posts, and code. This open approach allows external researchers and engineers to build on the company's work, fostering innovation and creating a culture of collaboration. By sharing knowledge and resources, Thinking Machines Lab aims to accelerate AI development and make these technologies more widely understood and accessible.

AI Safety and Ethics: Thinking Machines Lab is committed to ensuring the safe and ethical deployment of AI. The company places a strong emphasis on maintaining high safety standards for its AI systems, focusing on preventing misuse while maximising the freedom for users. The company also plays a role in promoting best practices for AI development by sharing guidelines and frameworks for safe AI construction. By engaging in real-world testing and sharing its findings, Thinking Machines Lab contributes to the broader AI community's efforts to develop systems that are both effective and ethically sound.

AI for Industry-Specific Applications: Thinking Machines Lab’s products are designed to serve a wide range of industries, from healthcare to finance and manufacturing. By developing AI systems that can be fine-tuned for specific industries, the company helps businesses unlock the full potential of AI in their operations. Whether for automating routine tasks, improving customer service, or driving innovations in product development, Thinking Machines Lab’s solutions are aimed at enhancing efficiency, accuracy, and productivity in various sectors.