Skip to content

Add MiniMax-M3 and make it the default model for the MiniMax provider#301

Closed
octo-patch wants to merge 1 commit into
1jehuang:masterfrom
octo-patch:feature/upgrade-minimax-m3
Closed

Add MiniMax-M3 and make it the default model for the MiniMax provider#301
octo-patch wants to merge 1 commit into
1jehuang:masterfrom
octo-patch:feature/upgrade-minimax-m3

Conversation

@octo-patch

Copy link
Copy Markdown

Summary

MiniMax's flagship text model is now MiniMax-M3. This PR pulls it into the minimax provider preset so users picking jcode login --provider minimax land on the current generation by default.

Changes

  • crates/jcode-provider-metadata/src/catalog.rs: bump MINIMAX_PROFILE.default_model from MiniMax-M2.7 to MiniMax-M3.
  • crates/jcode-base/src/provider_catalog.rs: rebuild the static model fallback for the minimax profile to expose MiniMax-M3, MiniMax-M2.7, and MiniMax-M2.7-highspeed. Drop the older M2.5 / M2.5-highspeed / M2.1 / M2.1-highspeed / M2 entries; live /v1/models will still surface anything else the user's account has.
  • crates/jcode-base/src/provider/openrouter_tests.rs: update the static-fallback regression to assert M3 is exposed (and replace the M2 assertion with the M3 one).
  • crates/jcode-base/src/live_tests.rs: switch the minimax live-test target model from M2.7 to M3.
  • OAUTH.md: refresh the default model hint in the MiniMax preset section.

Only the minimax provider's own list and default change; the other aggregators (opencode, cortecs, 302ai, alibaba-coding-plan) that mention older MiniMax model IDs in their static catalogs are left alone, since those reflect what each upstream actually exposes.

API base URL, env-file layout, China endpoint resolution, and the existing MiniMax-M2.7-highspeed entry are all untouched.

Notes

  • Heavily generated, in line with the project's stated workflow. CONTRIBUTING.md notes large/generated PRs may be used as a reference rather than merged as-is, which is fine here, the diff is intentionally small.
  • I did not run the full cargo test -p jcode-base suite locally because of build resource limits on the machine I have access to. The metadata crate (cargo check -p jcode-provider-metadata) builds clean. All edits are mechanical model-ID string replacements; the test changes line up 1:1 with the new static list.

Test plan

  • cargo test -p jcode-provider-metadata
  • cargo test -p jcode-base provider_catalog
  • cargo test -p jcode-base openrouter
  • Smoke: jcode login --provider minimax, confirm picker leads with MiniMax-M3 and that M2.7 / M2.7-highspeed are still reachable.

Adds the new flagship model MiniMax-M3 to the MiniMax provider's static
model list and promotes it to the default model hint. Keeps M2.7 and
M2.7-highspeed for users who still want them; drops the older
M2.5/M2.5-highspeed/M2.1/M2.1-highspeed/M2 entries from the static
catalog so the picker leads with the current generation.

Updates accompanying docs (OAUTH.md), the live-test target model, and
the static-model regression assertion to match the new list.

Co-Authored-By: Octopus <liyuan851277048@icloud.com>
@1jehuang 1jehuang added the triage:declined PR triage: out of scope / not planned label Jun 12, 2026
@1jehuang

Copy link
Copy Markdown
Owner

Thanks for the effort, but this is out of scope for jcode right now and the repo no longer accepts external PRs. Closing.

@1jehuang 1jehuang closed this Jun 12, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

triage:declined PR triage: out of scope / not planned

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants