Zed on Mingles Router

Run Zed on Mingles Router

Zed has a first-class openai_compatible provider. The one thing it insists on is a context length, and it fails unhelpfully without it.

Pick your model and paste your key — every command and value on this page updates live.

Endpoint: https://router.mingles.ai/v1 · key stays in your browser · no key entered? commands show <your-key>.

Add the provider

1. Edit settings.json

Open the command palette and run zed: open settings, then add:

settings.json
{
  "language_models": {
    "openai_compatible": {
      "mingles": {
        "api_url": "https://router.mingles.ai/v1",
        "available_models": [
          {
            "name": "deepseek-ai/DeepSeek-V4-Flash-0731",
            "display_name": "Mingles — deepseek-ai/DeepSeek-V4-Flash-0731",
            "max_tokens": 128000
          }
        ]
      }
    }
  }
}

max_tokens is required and is the context window, not the reply length. Omit it and the provider will not load.

2. Add the key — not in settings.json

Zed reads the key from the provider settings UI or from an environment variable named after the provider key you chose above:

shell
export MINGLES_API_KEY="<your-key>"

3. Pick it in the agent panel

The model appears in the model selector as "Mingles — {{model}}".

Exact values

api_url https://router.mingles.ai/v1
model name deepseek-ai/DeepSeek-V4-Flash-0731
max_tokens 128000

Switch model

Add more entries to available_models to switch between them from the selector — one block per model.

Before you file a bug: read the limits

Reasoning models spend the output budget on internal thinking, output is capped at 8192 tokens, there is no KV cache, and there are no built-in web tools. Most “it broke on Mingles Router” reports are one of these. See Model limits & behavior →

Frequently asked

The provider does not appear. +

Almost always a missing max_tokens, or the key not being set. The key belongs in the settings UI or the <PROVIDER>_API_KEY environment variable, not in settings.json.

Does Zed’s agent mode work with it? +

Yes — agent mode uses tool calling over chat completions, which is what this endpoint serves.