xTuring is an open-source library for personalising large language models. It covers the workflow from preparing a dataset to fine-tuning, including parameter-efficient LoRA fine-tuning, quantised INT8 training and inference, and running the tuned model, with an interactive playground for trying it. The documentation lists LLaMA, Falcon and GPT-J among the supported open-source models, with the full list under supported models and variants, and states three principles: ease of use, efficiency of compute and memory, and control over your own data and models.
It is developed by Stochastic under the Apache 2.0 licence. The GitHub repository was created in March 2023, has about 2,700 stars and 211 forks, and received a commit in September 2026, but the latest release on PyPI, 0.1.8, dates from September 2023, so installing from the package index gets a much older version than the repository head. Model support in the documentation reflects the 2023 generation of open models, so check that the current models you want to tune are supported before committing. No hosted service or pricing is mentioned.
Fine-tuning needs a GPU and the compute costs that come with it, which the library does not include. It suits developers who want to own the model and the data rather than call a hosted API, and a community Discord is linked from the repository.
