feat(minimax): add reasoning_split support for MiniMax M2 models

[darrenzeng2025]

Summary

This PR adds support for MiniMax's reasoning_split parameter to properly separate thinking content (cot) from response text when using MiniMax M2 models.

Problem

Issue #1320 reported that:

1. MiniMax official endpoint v1/text/chatcompletion_v2 is not supported
2. When using v1/chat/completions endpoint, thinking content (cot) appears in the text
3. Need to use reasoning_split:true to move cot to reasoning_details field

Solution

Added reasoning_split parameter support for MiniMax provider:

1. ModelConfig (pkg/config/config.go):

   • Added ReasoningSplit bool field for provider-specific options

2. openai_compat provider (pkg/providers/openai_compat/provider.go):

   • Added reasoningSplit field to Provider struct
   • Added WithReasoningSplit() option function
   • When enabled, adds reasoning_split: true to request body

3. HTTP Provider (pkg/providers/http_provider.go):

   • Added NewHTTPProviderWithOptions() constructor
   • Supports reasoningSplit parameter

4. Factory Provider (pkg/providers/factory_provider.go):

   • Separated MiniMax case to use NewHTTPProviderWithOptions()
   • Passes cfg.ReasoningSplit to provider

5. Migration (pkg/config/migration.go):

   • Added MiniMax to provider migration config

Usage

{
  "model_list": [
    {
      "model_name": "minimax-m2",
      "model": "minimax/MiniMax-M2.5",
      "api_key": "your-api-key",
      "reasoning_split": true
    }
  ]
}

When reasoning_split is enabled, MiniMax M2 models will return reasoning content in the reasoning_details field instead of embedding it within <think> tags in the content field.

Testing

• Code formatting verified with gofmt
• No syntax errors

Fixes #1320

[sipeed-bot]

@darrenzeng2025 Hi! This PR has had no activity for over 2 weeks, so I'm closing it for now to keep things tidy. If it's still relevant, feel free to reopen it anytime and we'll pick it back up.