Two-version jump from M2.1 (skipping intermediate M2.5). AA Intelligence Index 49.6 — top-tier Chinese open-weight reasoning at half the cost of Western frontier.
SI v3 uses weighted percentages from one Artificial Analysis configuration. Ranking requires all five benchmarks. Partial scores average only available weights. Scores are not comparable to SI v2 or a probability of reaching singularity.
| Benchmark | Score | Rank |
|---|---|---|
Arena EloArtificial Analysis Human preference ranking via blind comparisons | 1507 | #8 / 51 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 39.4% | #32 / 58 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 79.9% | #45 / 62 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 80.4% | #55 / 61 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 87.4% | #57 / 96 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 29.6% | #67 / 100 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisMiniMax-M2.7 | 29.6% |
Terminal-Bench 2.1 Artificial AnalysisMiniMax-M2.7 | 55.4% |
Terminal-Bench 4.0 Artificial AnalysisMiniMax-M2.7 | 0% |
SciCode Artificial AnalysisMiniMax-M2.7 | 50.1% |
AA-LCR Artificial AnalysisMiniMax-M2.7 | 78.3% |
CritPt Artificial AnalysisMiniMax-M2.7 | 0.6% |
ITBench SRE Artificial AnalysisMiniMax-M2.7 | 26.5% |
IFBench Artificial AnalysisMiniMax-M2.7 | 75.7% |