Cursor
Settings for using Echo Token Station models in the chat of the Cursor editor.
Cursor is a code editor with built-in AI features. With Cursor's "use your own API key" setting, you replace the OpenAI-format base URL with Echo.
Tested: No (as of 2026-09-27). The steps on this page are based on Cursor's official documentation and its forum.
Before you start
Cursor's official documentation lists the following restrictions.
- Your own API key can be used only with chat models. Tab completion (suggesting the rest of the code you are typing) keeps using Cursor's own models
- All requests go through Cursor's servers before they reach Echo. Cursor's terms also apply to what you send
We have not checked whether Echo's models work with the Agent feature.
Set up
Cursor's official documentation does not describe the setting for replacing the base URL. The setting names in the steps below are based on answers by Cursor staff in the Cursor forum and on the labels on screen.
- Open "Cursor Settings" and choose "Models"
- Enter your Echo API key in "OpenAI API Key"
- Turn on "Override OpenAI Base URL" and enter
https://token.echohub.io/v1 - Add the ID of the model you want to use (for example
deepseek-flash) to the model list. You can find model IDs in the model list - Enable the key and approve it on the confirmation screen
When "Override OpenAI Base URL" is on, every call that Cursor treats as an OpenAI model is sent to Echo (as reported in the Cursor forum). Turn it off again when you are not using Echo.
Claude models can also be called with this setting, but because the request is converted from the OpenAI format, structured output (specifying a JSON shape with response_format) has no effect.
Check that it works
Choose the model you added in the chat box and ask a simple question. If an answer comes back and the call appears in 「呼び出し履歴」 (Call history) in the console, you are done.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
| 404 with an HTML page instead of JSON | The Base URL does not end with /v1 | Change it to https://token.echohub.io/v1 |
401 and Invalid API key | The key is wrong, stopped or expired, or has used up its spending limit | Check the key's status in the console |
403 and Insufficient balance | Insufficient balance | Top up |
404 and Invalid URL | The Base URL has /v1 twice, or goes on to /chat/completions | Change it to https://token.echohub.io/v1 |
503 and The model … does not exist or is not available | The model ID is wrong, or Echo does not offer the model | Check the model ID |
The list of error bodies is in Errors and fixes.
Sources
| What | Source | Checked |
|---|---|---|
| Your own key works only with chat models, Tab completion uses Cursor's models, requests go through Cursor's servers | https://cursor.com/docs/settings/api-keys | 2026-09-26 |
| "Override OpenAI Base URL" under "Models" in "Cursor Settings", the report that the override applies to all OpenAI models | https://forum.cursor.com/t/openai-api-and-override-base-url-values/148140 | 2026-09-26 |
| HTTP statuses and error bodies on Echo's side, structured output not working when a Claude model is called in the OpenAI format | Measured in Echo's test environment (without Cursor) | 2026-09-27 |