Hey @ripe-winter-55813 Thanks for joining I am working through your issue on using minimax-m2.7 on OH locally

Hello, thanks for replying and helping me! I also tried default minimax configuration, but it didnt work either. Should i give the docker logs?

I used OpenHands to work through this. Can you try using prefix

openai/MiniMax-M2.7

or

anthropic/MiniMax-M2.7

I know it seems weird but OpenHands shared with me:

So:

• Yes, it should bypass the current LiteLLM provider-not-recognized error

• No, it does not guarantee everything else will work perfectly afterward

Best mental model

Think of it like this:

•

minimax/...

= “use LiteLLM’s native MiniMax handler”

•

openai/...

= “use LiteLLM’s OpenAI-style handler”

•

anthropic/...

= “use LiteLLM’s Anthropic-style handler”

Since MiniMax says it supports OpenAI-compatible and Anthropic-compatible APIs, those latter two are often the safer route in OpenHands.

So what should the user try?

First try the OpenAI-compatible route:

• Custom Model:

openai/MiniMax-M2.7

• Base URL:

<https://api.minimax.io/v1>

If that fails with a real API error, try Anthropic-compatible:

• Custom Model:

anthropic/MiniMax-M2.7

• Base URL:

<https://api.minimax.io/anthropic>

Short answer

Yes — using

openai/

or

anthropic/

can bypass this LiteLLM error, because those prefixes tell LiteLLM which compatibility adapter to use. They don’t claim the model is literally from OpenAI or Anthropic; they just select the request format.

Sorry for taking so much time to reply, i was trying those 2 options and restarted docker. For me OpenAI compatible route worked, Antrophic said i was out of credits (which is not possible). Thank you very much for help!