| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-06-01 | 750 Bytes | |
| v0.8.0 source code.tar.gz | 2026-06-01 | 46.1 MB | |
| v0.8.0 source code.zip | 2026-06-01 | 46.3 MB | |
| Totals: 3 Items | 92.4 MB | 0 | |
Changed
- MiniMax template updated from M2.7 to M3: All model slots now point to
MiniMax-M3(opus, sonnet, haiku, small/fast). M3 brings a 1M-token context window, frontier-level coding and agentic scores, native image/video input, and toggleable thinking. - Removed auto-compaction override from the MiniMax template. M3's 1M context window makes the 128K override unnecessary.
- MiniMax template
MAX_OUTPUT_TOKENSraised from 64K to 512K. - MiniMax template
CLAUDE_CODE_EFFORT_LEVELset tomax. - MiniMax template description updated to reflect M3 capabilities.