First reasoning model, uses chain-of-thought at inference time to solve hard problems
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 19 models released within six months of it, across 5 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
ARC-C Grade-school science questions requiring reasoning | 97.8% | #11 / 40 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 91.8% | #12 / 54 |
MATH Competition-level mathematics problems | 96.4% | #13 / 50 |
HumanEval Coding ability - generating correct Python functions | 92.4% | #26 / 50 |
Arena Elo Human preference ranking via blind comparisons | 1350 | #36 / 51 |
SWE-bench Real-world GitHub issue resolution | 48.9% | #37 / 39 |
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 0.8% | #40 / 40 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 12.9% | #52 / 58 |
GPQA PhD-level science questions even experts struggle with | 78% | #73 / 96 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 7% | #91 / 100 |
| Benchmark / source | Score |
|---|---|
HLE Artificial Analysiso1 | 7% |
AA-LCR Artificial Analysiso1 | 65% |
CritPt Artificial Analysiso1 | 0.3% |
IFBench Artificial Analysiso1 | 70.3% |