Anthropic's Claude 5 flagship. New ARC-AGI-2 state of the art at 97.1%, and the first model to post a meaningful ARC-AGI-3 score (30.2%).
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 76 models released within six months of it, across 4 benchmarks where scores spread out.
| Benchmark | Score | Rank |
|---|---|---|
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 91.6% | #2 / 61 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 89.9% | #2 / 51 |
AA Coding (Historical) Archived release-era coding composite, excluded from current rankings because comparable current-series coverage is unavailable | 78% | #2 / 22 |
AA Intelligence (Historical) Archived release-era composite, not comparable to the current versioned Intelligence Index used for rankings | 60.7% | #3 / 22 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 89% | #4 / 62 |
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 97.1% | #5 / 40 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 54.9% | #6 / 102 |
aaAgentic | 55.3% | #8 / 16 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 93.2% | #14 / 96 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisClaude Opus 5 (Max) | 54.9% |
MMMU-Pro Artificial AnalysisClaude Opus 5 (Max) | 84.7% |
Terminal-Bench 2.1 Artificial AnalysisClaude Opus 5 (Max) | 89.1% |
Terminal-Bench 4.0 Artificial AnalysisClaude Opus 5 (Max) | 49% |
SciCode Artificial AnalysisClaude Opus 5 (Max) | 56.4% |
AA-LCR Artificial AnalysisClaude Opus 5 (Max) | 79.3% |
CritPt Artificial AnalysisClaude Opus 5 (Max) | 29.1% |