Skip to content

feat(provider): add MiniMax-M2.7 (Global + CN)#2363

Open
kapelame wants to merge 1 commit into
logancyang:masterfrom
kapelame:add-minimax-m27
Open

feat(provider): add MiniMax-M2.7 (Global + CN)#2363
kapelame wants to merge 1 commit into
logancyang:masterfrom
kapelame:add-minimax-m27

Conversation

@kapelame

@kapelame kapelame commented Apr 21, 2026

Copy link
Copy Markdown

Adds MiniMax as a built-in OpenAI-compatible provider with MiniMax-M2.7 selectable from the model picker. Mirrors the SiliconFlow pattern.

Two regional providers since URLs and API keys differ:

format, lint, test (1904/1904) all pass.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 27e22c3141

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

Comment on lines +403 to +407
[ChatModelProviders.MINIMAX]: {
modelName: modelName,
apiKey: await getDecryptedKey(customModel.apiKey || settings.minimaxApiKey),
configuration: {
baseURL: customModel.baseUrl || ProviderInfo[ChatModelProviders.MINIMAX].host,

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Add MiniMax to sampling parameter allow-lists

This new MiniMax branch wires the provider through ChatOpenAI, but getProviderSpecificParams() is not updated to treat MINIMAX as supporting topP/frequencyPenalty, so any user-configured sampling overrides are silently dropped for MiniMax models while they still work for similar OpenAI-compatible providers like SiliconFlow and DeepSeek. This causes unexpected generation behavior whenever users tune those fields on a MiniMax custom model.

Useful? React with 👍 / 👎.

@kapelame kapelame changed the title feat(provider): add MiniMax-M2.7 as a built-in chat model feat(provider): add MiniMax-M2.7 (Global + CN) Apr 21, 2026
@logancyang

Copy link
Copy Markdown
Owner

Thanks for the contribution! We're planning to move provider integrations to a unified gateway approach soon, which may make individual provider PRs unnecessary. Putting this on hold until that direction settles. Will revisit then.

@logancyang logancyang added the hold Blocked by other dependencies label May 9, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

hold Blocked by other dependencies

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants