Simon Willison's CORS Chat provides a browser interface for talking directly to a configurable model endpoint. The endpoint must support OpenAI's Responses API format and allow the browser's cross-origin requests through CORS headers. Willison reports testing it successfully with LM Studio and OpenRouter. Project announcement
The interface stores conversation history in the browser and supports exporting it as JSON. That makes the interface and saved history independent of a particular chat service, while inferenceRunning a trained model to get answers — the phase where AI is actually used, as opposed to trained.Full definition → still happens at the selected endpoint. Choosing a local server or a remote provider therefore remains a separate decision from where the history is stored.
Terms in this piece · Glossary
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
streaming — Sending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.
