Skip to content

feat: add MiniMax as a new LLM provider#6723

Open
octo-patch wants to merge 1 commit into
ChatGPTNextWeb:mainfrom
octo-patch:feat/add-minimax-provider
Open

feat: add MiniMax as a new LLM provider#6723
octo-patch wants to merge 1 commit into
ChatGPTNextWeb:mainfrom
octo-patch:feat/add-minimax-provider

Conversation

@octo-patch

Copy link
Copy Markdown

Summary

Add MiniMax as a new LLM provider for NextChat. MiniMax offers OpenAI-compatible API with powerful models including the MiniMax-M2.5 series with up to 204K context window.

Changes

  • New files: app/api/minimax.ts (server-side API handler), app/client/platforms/minimax.ts (client platform)
  • Constants: Added MINIMAX_BASE_URL, MiniMax object, ServiceProvider.MiniMax, ModelProvider.MiniMax, ApiPath.MiniMax, and model definitions (MiniMax-M1/M2/M2.5 series)
  • Server config: MINIMAX_API_KEY and MINIMAX_URL environment variables
  • Access store: minimaxUrl, minimaxApiKey state and isValidMiniMax() validation
  • Auth: Server-side API key injection for MiniMax provider
  • Client API: MiniMaxApi class registration and API key resolution
  • Route handler: MiniMax route in the dynamic provider handler
  • Settings UI: MiniMax endpoint and API key configuration fields
  • Locales: English and Chinese translations
  • Documentation: Updated README.md and README_CN.md with MiniMax env vars

Special handling

  • MiniMax API requires temperature in (0.0, 1.0] — zero values are automatically clamped to 0.01
  • Supports streaming with thinking/reasoning content pass-through

Supported Models

  • MiniMax-M1, MiniMax-M1-80k
  • MiniMax-M2, MiniMax-M2-80k
  • MiniMax-M2.5, MiniMax-M2.5-80k

Test Plan

  • Verify TypeScript compilation passes (tsc --noEmit)
  • Test MiniMax provider selection in Settings UI
  • Test chat with MiniMax models using API key
  • Verify streaming responses work correctly
  • Verify non-streaming responses work correctly
Add MiniMax (https://api.minimax.io) as a new LLM provider with support
for MiniMax-M1, M2, and M2.5 series models. MiniMax provides an
OpenAI-compatible API with up to 204K context window.

Changes include:
- New client platform implementation (app/client/platforms/minimax.ts)
- New server-side API handler (app/api/minimax.ts)
- Provider constants, enums, and model definitions
- Server config for MINIMAX_API_KEY and MINIMAX_URL env vars
- Access store with validation and API key management
- Settings UI with endpoint and API key configuration
- Locale strings for English and Chinese
- Route handler registration
- README documentation for both EN and CN

Note: MiniMax API requires temperature in (0.0, 1.0], handled
automatically by clamping zero values to 0.01.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant