Support requesty AI provider as openrouter alternative

[jeroenev]

Requesty has some nice additional features, would be nice if it was added to Zed.
Since it's compatible with openrouter, it should be as simple as allowing us to change the openrouter endpoint in zed, like you can for openAI

[mkvalor]

This is the one missing feature preventing me from trying Zed. I suspect there may be others with the same opinion.

[PopCat19]

@jeroenev @mkvalor
On the latest version, you'll have to set OpenAI's custom API URL (e.g. https://router.requesty.ai/v1) and its API key on:
Agent Settings: LLM Providers: OpenAI
(don't forget to press enter/return to register it on input field; it should have checkmarks).

Then, manually add the model in settings.json like (it should already appear at top/head, if API is configured):

  "language_models": {
    "openai": {
      "api_url": "https://router.requesty.ai/v1", // API URL (will automatically be set from GUI)
      "available_models": [
        {
          "name": "coding/claude-4-sonnet", // aka `model=`
          "display_name": "Claude 4 Sonnet (coding)", // any name you want to call it
          "max_tokens": 128000
        }
      ]
    }
  },

You can find what (model=) name to copy in the Implementation section after you Select a model in: https://app.requesty.ai/getting-started

Docs Reference: https://zed.dev/docs/ai/configuration#openai-api-compatible
Issue Reference: #34327
Discussion for Auto Model Fetching: #34151

[jeroenev]

wasn't aware you could add the models you want to use when using the OpenAI Provider. Thanks a lot!

[mkvalor]

Thanks for making this clear!