Skip to content

feat(providers): refresh MiniMax model and endpoint coverage#665

Merged
kevincodex1 merged 1 commit into
Gitlawb:mainfrom
octo-patch:octo/20260713-minimax-m3-m27-provider-coverage-recvoLj0pEg0D3
Jul 13, 2026
Merged

feat(providers): refresh MiniMax model and endpoint coverage#665
kevincodex1 merged 1 commit into
Gitlawb:mainfrom
octo-patch:octo/20260713-minimax-m3-m27-provider-coverage-recvoLj0pEg0D3

Conversation

@octo-patch

@octo-patch octo-patch commented Jul 13, 2026

Copy link
Copy Markdown
Contributor

Supersedes #611 with a clean history based on the latest main branch.

Summary

  • Add MiniMax-M2.7 to the global and China provider model catalogs.
  • Record the current context, reasoning, tool-use, and input modality metadata for MiniMax-M3 and MiniMax-M2.7.
  • Enable image input for MiniMax-M3 and inline thinking parsing for M2.x models on OpenAI-compatible profiles.
  • Document the global and China Anthropic-compatible presets and OpenAI-compatible profile configuration.
  • Verify the final request URLs for both regions and both compatible protocols.

Verification

  • go fmt ./...
  • go test -p 1 ./...
  • make build-all
  • go vet ./...
  • make lint
  • go run github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.12.2 run --enable-only unused,ineffassign,staticcheck --new-from-rev=origin/main ./...
  • go run golang.org/x/vuln/cmd/govulncheck@v1.3.0 ./...

Summary by CodeRabbit

  • New Features

    • Added support for MiniMax M3 image input.
    • Expanded MiniMax model listings with current capabilities, including context limits, reasoning, tool calling, and supported modalities.
    • Added support for MiniMax M2 and M2.7 thinking output and compatible streaming endpoints.
    • Improved handling for global and China-region MiniMax configurations.
  • Documentation

    • Added setup instructions for MiniMax API keys, presets, and OpenAI-compatible custom profiles.
    • Documented region-specific base URLs and API key configuration.

@coderabbitai

coderabbitai Bot commented Jul 13, 2026

Copy link
Copy Markdown

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: b316b51a-645f-4a3e-9d40-b61589ba4902

📥 Commits

Reviewing files that changed from the base of the PR and between 5a5b7fd and f4ad7df.

📒 Files selected for processing (7)
  • README.md
  • internal/modelregistry/vision.go
  • internal/modelregistry/vision_name_test.go
  • internal/providermodelcatalog/catalog.go
  • internal/providermodelcatalog/catalog_test.go
  • internal/providers/factory.go
  • internal/providers/factory_test.go

Walkthrough

MiniMax setup documentation now covers API keys, presets, regional custom profiles, and base URLs. MiniMax catalog metadata, provider routing, think-tag parsing, and MiniMax M3 vision fallback behavior are updated with corresponding tests.

Changes

MiniMax support

Layer / File(s) Summary
MiniMax setup instructions
README.md
Documents MiniMax API key variables, global and China presets, and OpenAI-compatible custom profiles with regional URLs.
MiniMax catalog capabilities
internal/providermodelcatalog/catalog.go, internal/providermodelcatalog/catalog_test.go
Adds detailed capabilities for MiniMax M3 and M2.7, updates provider model expectations, and verifies catalog results do not alias mutable state.
Provider routing and think tags
internal/providers/factory.go, internal/providers/factory_test.go
Enables think-tag handling for MiniMax M2 models and tests global and China-compatible streaming endpoints.
MiniMax M3 vision fallback
internal/modelregistry/vision.go, internal/modelregistry/vision_name_test.go
Recognizes MiniMax M3 as vision-capable when resolved through name-based fallback logic.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Possibly related PRs

  • Gitlawb/zero#200: Introduces streaming logic related to the MiniMax think-tag parsing path.
  • Gitlawb/zero#398: Modifies the MiniMax curated model catalog and provider definitions.

Suggested reviewers: vasanthdev2004, gnanam1990, jatmn, kevincodex1

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name Status Explanation Resolution
Docstring Coverage ⚠️ Warning Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Write docstrings for the functions missing them to satisfy the coverage threshold.
✅ Passed checks (4 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The title accurately summarizes the main MiniMax model and endpoint coverage refresh in the changeset.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests

Warning

Tools execution failed with the following error:

Failed to run tools: 13 INTERNAL: Received RST_STREAM with code 2 (Internal server error)


Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@Vasanthdev2004 Vasanthdev2004 left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Reviewed #665 against the prior #611 concerns. This cleanly supersedes #611: it leaves the MiniMax base URLs at /anthropic (the provider descriptor file is untouched), so there's no /v1 doubling, and the new endpoint test asserts the resolved request URL for both regions and both compatible protocols — exactly the regression test that was asked for. Build, tests on the touched packages, vet, and gofmt are all clean, and the merge applies cleanly.

I also checked for the attribution pattern from the earlier preset PR: no partner/rebate/affiliate header, no recommended-tier elevation of MiniMax, and no hardcoded signing key or official-source trust anchor — the only new constant is a docs URL used as the model Source field. #611 is closed and should stay closed in favor of this one. The missing linked approved issue is the same gap that was waived on #611; I'm not blocking on it for a model refresh of an existing provider.

One minor note: the new minimax-m2 think-tag marker in the provider wiring also matches MiniMax-M2.1, which the catalog doesn't mark as a reasoning model — harmless in practice since stripping tags a model never emits is a no-op, but could be tightened to minimax-m2.7 later if M2.1 is definitively non-reasoning.

@kevincodex1 kevincodex1 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

looks great. thanks for contribution. please join discord

@kevincodex1 kevincodex1 merged commit fa3052a into Gitlawb:main Jul 13, 2026
7 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants