feat: upgrade MiniMax default model to M3#37
Open
octo-patch wants to merge 2 commits into
Open
Conversation
- Add MiniMax-M2.7 and MiniMax-M2.7-highspeed to model catalog - Set MiniMax-M2.7 as default model across all configurations - Update synthetic provider default to hf:MiniMaxAI/MiniMax-M2.7 - Add M2.7 prefixes to live model filter - Update onboard auth configs and wizard docs - Update English and Chinese provider documentation - Keep all previous models (M2.5, M2.5-highspeed, M2.5-Lightning) as alternatives - Update related unit tests
- Add MiniMax-M3 as the new default model (512K context, 128K max output, image input) - Update MINIMAX_HOSTED_MODEL_ID, SYNTHETIC_DEFAULT_MODEL_ID, and MiniMax portal default - Update DEFAULT_MINIMAX_CONTEXT_WINDOW=512000 and DEFAULT_MINIMAX_MAX_TOKENS=128000 - Keep MiniMax-M2.7 as the previous flagship; remove M2.5/M2.5-highspeed/M2.5-Lightning from default catalog - Update live model filter prefixes (m3, m2.7) - Update bench-model, MiniMax OAuth plugin, onboard auth choice and config defaults - Refresh synthetic-provider catalog, model-providers docs, configuration examples, faq, wizard reference - Update zh-CN provider docs to mirror English changes - Update related unit tests to expect M3 as default
octo-patch
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Upgrade MiniMax model configuration to use the latest M3 flagship model as the new default, with M2.7 retained as the previous flagship.
Changes
MiniMax-M3to the model catalog and set it as the new default across all configurations:MINIMAX_HOSTED_MODEL_ID→MiniMax-M3SYNTHETIC_DEFAULT_MODEL_ID→hf:MiniMaxAI/MiniMax-M3buildMinimaxProvider()andbuildMinimaxPortalProvider()defaultsminimax-m3)reasoning: trueMiniMax-M2.7as the previous flagship in the catalog andMiniMax-VL-01for visionMiniMax-M2.5,MiniMax-M2.5-highspeed,MiniMax-M2.5-Lightning) from the default catalog and provider definitions24 files changed across source, tests, and documentation.
Why
MiniMax-M3 is the latest flagship model with enhanced reasoning, coding, and image-input capabilities, larger context (512K vs 192K), and higher max output (128K vs 65K). M3 supersedes both M2.7 and M2.5 as the recommended default. M2.7 is retained for backward compatibility and as a fast tier.
Testing
lmstudio/minimax-m2.5-gs32(quantized local model) intentionally left untouched