Unified reasoning and chat, 400K context, 94.6% on AIME 2025
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.
Compared with 28 models released within six months of it, across 9 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
HellaSwag Common sense reasoning about everyday situations | 97.2% | #4 / 36 |
ARC-C Grade-school science questions requiring reasoning | 98.1% | #9 / 40 |
HumanEval Coding ability - generating correct Python functions | 94.8% | #11 / 50 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 91.2% | #15 / 54 |
SWE-bench Real-world GitHub issue resolution | 74.9% | #16 / 39 |
MATH Competition-level mathematics problems | 94.6% | #18 / 50 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 86.6% | #19 / 62 |
Arena Elo Human preference ranking via blind comparisons | 1398 | #30 / 51 |
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 18.3% | #32 / 40 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 78.9% | #39 / 51 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 33.3% | #40 / 58 |
GPQA PhD-level science questions even experts struggle with | 88.4% | #48 / 96 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 82.2% | #53 / 61 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 21.5% | #78 / 100 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisGPT-5 mini (High) | 21.5% |
MMMU-Pro Artificial AnalysisGPT-5 mini (High) | 70.1% |
Terminal-Bench 2.1 Artificial AnalysisGPT-5 mini (High) | 3.7% |
Terminal-Bench 4.0 Artificial AnalysisGPT-5 mini (High) | 0% |
SciCode Artificial AnalysisGPT-5 mini (High) | 39% |
AA-LCR Artificial AnalysisGPT-5 mini (High) | 72.3% |
CritPt Artificial AnalysisGPT-5 mini (High) | 0% |
IFBench Artificial AnalysisGPT-5 mini (High) | 75.4% |