Frontier-tier successor to M2.7. AA Intelligence Index 54.7, GPQA 92.9, Arena 1668 — continues the open-weight cost-disruption pattern.
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 | 1668 | #2 / 51 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 92.9% | #16 / 96 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 42.4% | #28 / 58 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 81.2% | #35 / 51 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 82.2% | #37 / 62 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 84.2% | #48 / 61 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 39% | #49 / 102 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisMiniMax-M3 | 39% |
MMMU-Pro Artificial AnalysisMiniMax-M3 | 78.6% |
Terminal-Bench 2.1 Artificial AnalysisMiniMax-M3 | 65.2% |
Terminal-Bench 4.0 Artificial AnalysisMiniMax-M3 | 2% |
SciCode Artificial AnalysisMiniMax-M3 | 47.1% |
AA-LCR Artificial AnalysisMiniMax-M3 | 83% |
CritPt Artificial AnalysisMiniMax-M3 | 3.7% |
IFBench Artificial AnalysisMiniMax-M3 | 82.9% |