Proprietary reasoning model with text and image input. Independent evaluations and pricing are reported by Artificial Analysis.
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 3 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 46.5% | #18 / 95 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisStep 5 Preview | 46.5% |
MMMU-Pro Artificial AnalysisStep 5 Preview | 76.4% |
Terminal-Bench 4.0 Artificial AnalysisStep 5 Preview | 33.3% |
SciCode Artificial AnalysisStep 5 Preview | 58.9% |
AA-LCR Artificial AnalysisStep 5 Preview | 88.3% |
CritPt Artificial AnalysisStep 5 Preview | 20.9% |
ITBench SRE Artificial AnalysisStep 5 Preview | 55.6% |
GPQA Diamond Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 93.5% |
HLE Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 46.5% |
AA-LCR 1.1 Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 88.3% |
CritPt Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 20.9% |
DeepSWE 1.1 Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent; SWE-agent; temperature 1.0, top_p 0.95 | 67.7% |
Terminal-Bench 2.1 Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 85% |
Terminal-Bench 4.0 Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 33.3% |
CyberGym Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 84.7% |
SciCode Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 58.9% |
ProgramBench Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 80.5% |
StepCodeBench Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent; avg@4 | 49% |
Agents' Last Exam (ALE-CLI) Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 29.5% |
GDPval-AA 2.1 Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 1566 Elo |
FrontierFinance Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 66.4% |
DRACO Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 83.3% |
FinStepBench LiveSearch Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 74.5% |
FinStepBench CorporateValuation Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 60.6% |
FinStepBench DeepResearch Provider reportStep 5 Preview high effort; provider evaluation, harness-dependent | 55.8% |