MiniMax M3.1 Flash Card - SWE Refactor Bench: leaderboard
MiniMax M3.1 Flash official-weight card; native publisher-owned observation only. - **SWE Refactor Bench:** Claude Code 2.1.220, 200K context; otherwise official setup. Reasoning effort,attempt count and unreported settings remain unspecified; recommended inference settings are not asserted evaluation settings. Comparative cells lack explicit per-cell evaluator provenance and are withheld. GDP.pdfAA,ALEofficial andOSWorldofficial mirrors are separately identified and not duplicated. CyberGym82.2 is compatible with release-prose82% at published whole-percent precision; no cross-source run fusion is inferred.
Metric: Reported score (0–100). Source: huggingface.co. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | MiniMax-M3.1-Flash | 23 | #52 |
Interactive version: theaggregate.ai/benchmark?slug=minimax-m3-1-flash-card-swe-refactor-bench · How It Works · Data refreshed daily, snapshot 2026-11-05.