Skip to content

feat(core,web,docs): add official MiniMax M3 preset - #167

Draft
Youhai020616 wants to merge 4 commits into
mainfrom
feat/minimax-m3
Draft

feat(core,web,docs): add official MiniMax M3 preset#167
Youhai020616 wants to merge 4 commits into
mainfrom
feat/minimax-m3

Conversation

@Youhai020616

@Youhai020616 Youhai020616 commented Aug 3, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Adds the official minimax-token-plan provider group with one direct MiniMax-M3 preset.
  • Pins AgentHub client type minimax-m3 and the official endpoint https://api.minimax.io/v1.
  • Uses MINIMAX_API_KEY / MINIMAX_BASE_URL credential fallback and includes the official MiniMax provider glyph.
  • Updates core and web tests, bilingual model/configuration docs, and the unreleased changelog.

Scope

This PR supports only MiniMax-M3 with a 1,000,000-token context window and vision input.

  • MiniMax-M2.7 and MiniMax-M2.7-highspeed are not included.
  • M3 pricing is intentionally omitted because MiniMax doubles all rates above 512K input tokens and the catalog cannot represent tiered pricing.
  • Env resolution mirrors AgentHub exact routing: a non-M3 model remains invalid even when client_type = "minimax-m3" is supplied.
  • The existing local/vLLM documentation and current main model catalog updates are preserved.

Validation

  • pnpm format:check
  • pnpm typecheck
  • pnpm test
  • pnpm build
  • git diff --check
  • Core catalog tests: 14/14 passed
  • Web model-grouping tests: 18/18 passed

The first full test run hit an existing timing-sensitive exec-session assertion under parallel load; the test passed in isolation and the complete suite passed on rerun. The full build passed; the optional global CLI link was skipped because the local pnpm global bin directory is not in PATH.

Draft dependency blocker

Keep this PR in draft until:

  1. Merge AgentHub PR Add MiniMax M3 Responses API support agenthub#167.
  2. Release @prismshadow/agenthub 0.4.2.
  3. Bump PenguinHarness core/CLI dependencies and the lockfile from ^0.4.1 to ^0.4.2.
  4. Re-run validation and then mark this PR ready for review.

Youhai020616 and others added 4 commits August 3, 2026 20:13
…client

New minimax-token-plan provider group with MiniMax-M3 / M2.7 /
M2.7-highspeed catalog entries pinning the shared minimax-m3 protocol
and direct endpoint, MINIMAX_* env resolution mirroring AgentHub's
exact-match routing, a MiniMax provider glyph, and bilingual model and
configuration docs (including the M2.7 none-to-low thinking mapping).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Replace the placeholder M letterform with MiniMax's official
stream-lines mark, flattened to currentColor like the other vendor
brand marks (recognition purposes, evenodd fill preserved).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@Youhai020616 Youhai020616 changed the title feat(core,web,docs): add MiniMax M-series presets via the minimax-m3 client feat(core,web,docs): add official MiniMax M3 preset Aug 10, 2026
{
modelId: "MiniMax-M3",
displayName: "MiniMax M3",
provider: "minimax-token-plan",

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.

这个不需要分 Token plan

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.

2 participants