Strongest agentic model at launch, sustained multi-hour autonomous coding
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 25 models released within six months of it, across 10 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
ARC-C Grade-school science questions requiring reasoning | 97.2% | #16 / 40 |
HumanEval Coding ability - generating correct Python functions | 93.2% | #21 / 50 |
SWE-bench Real-world GitHub issue resolution | 72.5% | #22 / 39 |
HellaSwag Common sense reasoning about everyday situations | 91.2% | #23 / 36 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 88.8% | #26 / 54 |
MATH Competition-level mathematics problems | 84.6% | #31 / 50 |
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 8.6% | #36 / 40 |
Arena Elo Human preference ranking via blind comparisons | 1342 | #37 / 51 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 86.2% | #39 / 61 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 31.1% | #41 / 58 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 73.3% | #44 / 51 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 70.2% | #54 / 62 |
GPQA PhD-level science questions even experts struggle with | 79.6% | #72 / 96 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 12.3% | #86 / 100 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisClaude 4 Opus (Reasoning) | 12.3% |
AA-LCR Artificial AnalysisClaude 4 Opus (Reasoning) | 69.3% |
IFBench Artificial AnalysisClaude 4 Opus (Reasoning) | 53.7% |