ModelBump¶
Dependabot for LLM models. ModelBump finds the AI models your code depends on and warns you before a provider retires them. Later versions will also test the old model against its replacement, repair prompts that regress, and open the upgrade pull request for you.
v0.3.0: the scanner, the runner and the statistics engine
The first stage, Scan, is out as a command-line tool and a GitHub Action:
pip install modelbump
modelbump scan .
See the scanner for the options and the Action. Two building blocks for the next stages are out too: the model runner (modelbump ask) and the statistics engine (modelbump stats), which tells a real regression from normal LLM randomness. The other stages are still being built.
Where to look¶
| Section | What you'll find |
|---|---|
| Architecture | How the system is structured and why |
| Decisions (ADRs) | Every significant technical decision, with the alternatives we rejected |
| Engineering challenges | Problems we hit while building, and how we solved them |
| Dev journal | A short daily log of progress |
| Research | Research questions, experiment protocols and results |