Lamini is an enterprise platform for fine-tuning and deploying large language models, and this repository is its official open-source Python client and SDK. The package has around 2,500 GitHub stars.
The lamini Python package connects your code to the Lamini API: install it with pip, drop your API key into a small config file, and you can fine-tune models on your own data and run inference through the platform. The client handles authentication, request plumbing and response parsing, so training jobs can be driven from notebooks and scripts. Lamini positions itself around helping companies create their own generative AI โ training models that absorb proprietary documents and workflows rather than relying on a general-purpose chatbot. Typical uses include support assistants trained on internal knowledge, extraction pipelines over company documents, and domain-specific copilots. The SDK is intentionally thin and kept current with the platform, and the company recommends staying on the latest stable release. Documentation for the package is published separately; this repo tracks the client itself rather than the hosted service.
The Python SDK is free and open source. The Lamini platform behind it is a commercial service with paid plans for training and inference, so costs depend on your usage of the hosted API; check Lamini’s site for current plan details.
Enterprise developers who want to fine-tune and serve custom LLMs through a managed platform using a lightweight Python SDK, and teams evaluating managed fine-tuning against self-hosting.