Minimax configuration

[aligzl]

Hi,
is there anyone tried minimax configuration? I made fallowing config but it doesnt work good as expected.

{
  "providers": {
    "minimax": {
    "baseUrl": "https://api.minimax.io/anthropic",
    "apiKey": "****",
    "api": "anthropic-messages",
      "models": [
        {
          "id": "minimax2.7",
          "name": "Minimax 2.7",
          "reasoning": true,
          "contextWindow": 205000,
          "maxTokens": 32000,
          "cost": {
            "input": 0.5,
            "output": 1.5,
            "cacheRead": 0,
            "cacheWrite": 0
          }
        }
      ]
    }
  }
}

any suggestions?
thanks in advance

[jeremymcs]

@aligzl we'll work on minimax and other models.

[aligzl]

Thx for the great work 👍

[trek-e]

Native MiniMax integration is now tracked in issue #4592. In the meantime, the Anthropic-compatible endpoint approach you tried (api: "anthropic-messages") is the right path — the main gaps are context window sizes and model enumeration, which are the specific items tracked in that issue.

For now, explicitly specifying your models list in the config (as you did) is the best workaround. If you see specific errors or degraded behavior with your setup, please file them on #4592 with the config and error output.