MiniMax M3.1 Flash
Announced on minimax.io
Benchmark results
| Benchmark | Score | Source check |
|---|---|---|
| Humanity's Last ExamNo tools · text-only subset | 43.4% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| Terminal-Bench 4.0Codex 0.149.1 harness | 38.1% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| AutomationBenchAutomationBench v1.0.6; 600 public tasks | 62.8% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| low effort; AutomationBench v1.0.6; Claude Code harness | 52.8% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| medium effort; AutomationBench v1.0.6; Claude Code harness | 57.7% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| high effort; AutomationBench v1.0.6; Claude Code harness | 59.2% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| xhigh effort; AutomationBench v1.0.6; Claude Code harness | 61.2% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| OSWorld 2.0Offline set, partial score · task release v2026.08.08 | 61.7% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| DeepSWE v1.1mini-SWE-agent 2.4.5 harness; 512K context | 73.7% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| high effort; mini-swe-agent harness | 68.1% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| xhigh effort; mini-swe-agent harness | 72.1% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
| CyberGymClaude Code 2.1.263 harness; 1M context | 82.2% | CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced. |
Sources
Reported by MiniMax
