Thinking Machines Lab
Mira Murati's frontier AI lab, whose first product Tinker is an API for fine-tuning open-weight language models.
Overview
Thinking Machines Lab is a frontier AI research company founded by former OpenAI CTO Mira Murati. In July 2025 it raised a record $2 billion seed round at a $12 billion valuation. Its first product, Tinker (launched October 2025), is an API that gives researchers low-level control over fine-tuning open-weight models. Reported talks to raise at a ~$50 billion valuation collapsed in early 2026, leaving the lab at its original $12 billion mark.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 3 Sep 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with Thinking Machines Lab
- Fine-tuning of open-weight models through Tinker, which Karpathy framed as keeping roughly 90% of algorithmic control while removing about 90% of the infrastructure pain
- LoRA-based tuning runs without standing up your own training infrastructure
- Results of the kind early academic adopters reported — Princeton's Goedel team matched full fine-tuning on a fraction of the data
- ⚠️ A waitlist rather than a signup: access is gated, and hands-on community sentiment is still thin as a result
- A researcher's tool rather than a consumer product — sophisticated for ML researchers, and it landed with less awe than the funding headlines implied
- ⚠️ An open question rather than a settled vendor: talks to raise at roughly $50 billion collapsed in early 2026, leaving the lab back at its $12 billion mark, amid senior departures including the reported firing of co-founder and CTO Barret Zoph and Meta poaching founding members
What this is for: A frontier research lab whose first shipped product, Tinker, is an API for fine-tuning open-weight language models with fine-grained control over data and algorithms.
Who this is for: ML engineers and researchers who want programmatic control over model fine-tuning rather than a black-box endpoint.
Availability: Paid, usage-based API access; onboarding is gated behind a waitlist. Founded by ex-OpenAI CTO Mira Murati.
What people say
Thinking Machines Lab drew unusual attention for a pre-product company on the strength of Mira Murati's name and a record $2 billion seed. When Tinker finally shipped in October 2025, the loudest praise came from Andrej Karpathy, who framed its design as retaining roughly 90% of algorithmic control while removing about 90% of infrastructure pain; early adopters at Berkeley, Princeton, Stanford, and Redwood Research reported strong results, with Princeton's Goedel team matching full fine-tuning using a fraction of the data. VentureBeat and others noted that Tinker is a sophisticated tool for ML researchers, not a consumer breakthrough, and it landed with less awe than the funding headlines suggested. It is also LoRA-focused and gated behind a waitlist, so hands-on community sentiment is still thin.
The harder skepticism is about the company, not the product. Reported talks to raise at roughly $50 billion collapsed in early 2026 because backers were unwilling to pay that price against a single shipped product, leaving the lab back at its $12 billion mark. That timing coincided with turmoil inside the company, including senior departures, the reported firing of co-founder and CTO Barret Zoph, and Meta poaching several founding members. The result is a well-regarded first product wrapped in an open question about whether the valuation and the team can hold.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Thinking Machines Lab's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Thinking Machines Lab unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Thinking Machines Lab directly →
Spotted something out of date? Suggest an update →
Alternatives to Thinking Machines Lab
More in LLMs & Chat
