Reasoning model for coding and professional work. Independent measurements track max effort.
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 58 models released within six months of it, across 4 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 90% | #10 / 39 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 47.9% | #13 / 95 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisGPT-6 Sol (Max) | 47.9% |
MMMU-Pro Artificial AnalysisGPT-6 Sol (Max) | 82.9% |
Terminal-Bench 4.0 Artificial AnalysisGPT-6 Sol (Max) | 43.9% |
SciCode Artificial AnalysisGPT-6 Sol (Max) | 57.6% |
AA-LCR Artificial AnalysisGPT-6 Sol (Max) | 83.7% |
CritPt Artificial AnalysisGPT-6 Sol (Max) | 30.9% |
ITBench SRE Artificial AnalysisGPT-6 Sol (Max) | 49.4% |
DeepSWE 1.1 Provider reportOpenAI research/API evaluation; effort specified per result; Max effort | 68.8% |
Agents' Last Exam V1 Provider reportOpenAI research/API evaluation; effort specified per result; Max effort | 56.4% |
AutomationBench 1.0.6 Provider reportOpenAI research/API evaluation; effort specified per result; Xhigh effort | 33.2% |
OSWorld 2.0 offline Provider reportOpenAI research/API evaluation; effort specified per result; Xhigh effort; partial reward; v2026.08.08 | 60.5% |