Source: simonwillison.net/2026/Nov/14/datasette-llm
datasette-llm is an intermediate plugin that configures LLMs for other Datasette plugins to use. Version 0.1a11 adds support for OpenAI-compatible endpoints, so the model does not have to come from a particular hosted provider. Any server that implements that protocol can be set up as the model backend.
Willison's example uses local models run through LM Studio, Ollama or llama.cpp, which he notes has just reached v1.0.0. The default model is set through plugins.datasette-llm.default_model as a JSON object with a model name, a model_name and an api_base such as http://localhost:1234/v1. Running datasette with uvx and the datasette-agent plugin then starts a local Datasette Agent instance powered by that model.
Why it matters if you build agents
Developers building agents on Datasette can point its plugins at a local or self-hosted model through a config entry, without writing custom client code, as long as the server exposes an OpenAI-compatible API. The example shows one configuration, so the api_base and model name need to match the server that is actually running.
Get stories like this every weekday
One short email with the day's agent tooling news.