Rankings / Unknown
Claude 4 Opus
released 2025-05-22
BenchAtlas Index
as of 2026-07-14
thinking
rank #197
11 families · 5 categories · high
no reasoning
rank #233
9 families · 5 categories · high
Benchmark evidence
26 results
agentic coding
| τ²-Bench | thinking | 73.4 % | independent | Artificial Analysis |
coding
| LiveCodeBench v6 implementation=artificial-analysis | thinking | 63.6 % | independent | Artificial Analysis |
| LiveCodeBench v6 implementation=artificial-analysis | no reasoning | 54.2 % | independent | Artificial Analysis |
| SciCode | no reasoning | 40.9 % | independent | Artificial Analysis |
| SciCode | thinking | 39.8 % | independent | Artificial Analysis |
external indices
| Intelligence Index v4.1 | thinking | 31.0 points | independent | Artificial Analysis |
| Intelligence Index v4.1 | no reasoning | 25.5 points | independent | Artificial Analysis |
| AA Math Index | thinking | 73.3 points | independent | Artificial Analysis |
| AA Math Index | no reasoning | 36.3 points | independent | Artificial Analysis |
knowledge science
| GPQA Diamond implementation=artificial-analysis | thinking | 79.6 % | independent | Artificial Analysis |
| GPQA Diamond implementation=artificial-analysis | no reasoning | 70.1 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | thinking | 11.7 % | independent | Artificial Analysis |
| Humanity's Last Exam implementation=artificial-analysis | no reasoning | 5.9 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | thinking | 87.3 % | independent | Artificial Analysis |
| MMLU-Pro implementation=artificial-analysis | no reasoning | 86.0 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | no reasoning | 36.0 % | independent | Artificial Analysis |
| AA-LCR | thinking | 33.7 % | independent | Artificial Analysis |
| IFBench | thinking | 53.7 % | independent | Artificial Analysis |
| IFBench | no reasoning | 43.3 % | independent | Artificial Analysis |
| MultiChallenge | thinking | 58.6 % 57.9–59.4 | independent | Scale Labs |
reasoning math
| AIME implementation=artificial-analysis | thinking | 75.7 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | thinking | 73.3 % | independent | Artificial Analysis |
| AIME implementation=artificial-analysis | no reasoning | 56.3 % | independent | Artificial Analysis |
| AIME year=2025 · implementation=artificial-analysis | no reasoning | 36.3 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | thinking | 98.2 % | independent | Artificial Analysis |
| MATH-500 implementation=artificial-analysis | no reasoning | 94.1 % | independent | Artificial Analysis |
Agent + model results
systems, not bare-model scores
| agent + model Artificial Analysis harness + Claude 4 Opus | Terminal-Bench Hard | 31.1 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.