Fine-tuningCommercial
Tinker
Managed distributed fine-tuning API from Thinking Machines Lab
App Infrastructure
About
Hosted fine-tuning API (LoRA and full fine-tune) for open-weight models, including large MoE models, from Mira Murati's Thinking Machines Lab. A managed-training-loop alternative to library-based tools like Axolotl or Unsloth — usage-based per-token pricing that varies by model.
Choose Tinker when…
- •You want a managed fine-tuning API instead of running training infra yourself
- •You need to fine-tune large open-weight MoE models without provisioning GPU clusters
- •You're comparing a hosted training loop against library-based options like Axolotl/Unsloth
Builder Slot
How do you adapt models to your domain?Optional for most stacks
Fine-tuning frameworks and platforms for training custom model adaptations with LoRA, QLoRA, or full fine-tuning
Dev Tools
Not applicable
App Infra
Optional
Hybrid
Optional
Alternatives to consider (2)
Pricing
Usage-basedPer-token, varies by model
Pulse
● No incidents in the last 90 days
Badge
Add to your GitHub README
[](https://www.aichitect.dev/tool/tinker)Explore the full AI landscape
See how Tinker fits into the bigger picture — browse all 207 tools and their relationships.