Former OpenAI CTO Mira Murati has officially launched her first product under her new startup, Thinking Machines Lab. The debut offering, called Tinker, is a free API designed to make it easier for developers and researchers to fine-tune large language models (LLMs).
Curious about how other leading AI tools compare for fine-tuning, generation, or automation? Explore top platforms like Claude, ChatGPT, Gemini, Mistral, and more in our full list:
Explore All Top AI Tools → earlyhow.com/tools
What Is the Tinker API Designed to Do?
Tinker allows users to fine-tune LLMs using simple, modular abstractions—without requiring deep ML engineering expertise. The platform is positioned as a developer-friendly tool that supports:
- Lightweight LLM customization
- Safe and reproducible model updates
- Tuning for security, alignment, and UX-driven behaviors
The API is currently in an early-access phase, with functionality tied to language-based prompting. According to the launch notes, Tinker works well with popular open-source models such as LLaMA 2, Mistral, and Gemma.
Who Is Tinker Built For?
Thinking Machines says Tinker is aimed at AI builders across multiple sectors—especially those who want to rapidly adapt models for internal use, product prototypes, or research environments.
It’s especially relevant for:
- Teams working on AI safety, reliability, and alignment
- Developers building customer-facing tools with LLMs
- Researchers experimenting with smaller or domain-specific models
By abstracting away many of the painful setup steps, Tinker makes fine-tuning more accessible and modular, opening the door to broader experimentation in AI systems.
Tinker Cookbook Adds Usability for Developers
Alongside the API, Murati’s team also launched the Tinker Cookbook—an open-source collection of usage patterns, recipes, and code snippets for real-world fine-tuning.
The Cookbook includes:
- Sample prompt schemas
- Pre-built configurations for common use cases
- Guides for training runs on Hugging Face, Google Colab, and local GPU environments
This complements Tinker’s vision of making LLM optimization easier, faster, and repeatable across both individual and team workflows.
Who’s Behind the Project?
Thinking Machines Lab was co-founded by Murati in early 2024. The founding team includes alumni from OpenAI, Google DeepMind, Meta AI, and Anthropic.
The company has not yet disclosed outside funding, but insiders expect a formal raise to follow the Tinker launch as usage grows.
According to early users, Tinker stands out because it strikes a balance between:
- Accessibility and simplicity
- Research-grade depth
- Community-driven modularity
Final Thoughts
Tinker represents a different approach to AI infrastructure—one focused on control, usability, and openness rather than proprietary black boxes.
It’s still early, but with Mira Murati at the helm and a clear focus on user-first design, Thinking Machines Lab may become a key player in the tooling layer of the AI ecosystem.
What’s your take on Tinker and Murati’s mission to simplify model fine-tuning?
Is open tooling the future of LLM development, or will proprietary APIs always dominate?
Let us know what you think in the comments below.
Follow EarlyHow.com for in-depth reviews of AI APIs, dev tools, and startup releases shaping the AI landscape.
