Tinker is Thinking Machines Lab's first public product, a Python-based API for fine-tuning large language models that handles distributed compute and infrastructure management so developers can focus on the training pipeline itself. It reached general availability in December 2025 and includes a sampling tool for testing a model's responses mid-training.
← All AI tools
Visit Website
Pricing
Usage-based API pricing
Best For
Distributed fine-tuning of large language models via API
