BenchLeader

Claude Opus 5 vs GPT-5.6 Sol

Verdict
  • Claude Opus 5 leads on quality: 70.0 vs 68.0.
  • Claude Opus 5 is stronger in coding, composite, human preference, maths, multimodal, reasoning.
  • GPT-5.6 Sol (high) is stronger in agents & tools, knowledge, long context, instruction following.
  • GPT-5.6 Sol (high) is 1.3× cheaper ($8.00 vs $10.00 per 1M blended).
MetricClaude Opus 5GPT-5.6 Sol (high)
BenchLeader Index70.068.0
Agents & tools score71.687.6
Coding score74.472.0
Composite score92.982.5
Human preference score78.2
Knowledge score70.573.7
Long context score66.267.4
Maths score67.1
Multimodal score69.466.5
Reasoning score71.162.9
Instruction following score67.6
Blended price $/M$10.00$8.00
Output speed52 tok/s59 tok/s
Time to first answer81.6 s31.4 s
Context window1M1M
GPQA Diamond92.9%
OTIS Mock AIME97.8%
SimpleBench80.6%
SciCode56.9%
WeirdML86.3%88.8%
Epoch Capabilities Index162.6
AA Intelligence Index50.742.5
IFBench69.2%
AA-LCR79.3%81.7%
MMMU-Pro84.7%81.8%
AA-Omniscience37.120.4
Terminal-Bench Hard62.1%
GPQA Diamond (AA)93.2%92.8%
Humanity's Last Exam (AA)54.9%46.0%
SciCode (AA)56.4%57.8%
τ²-Bench Telecom (AA)83.3%
LiveCodeBench89.0%
MMLU-Pro91.6%
IOI91.7%
LegalBench87.0%
CorpFin73.2%
TaxEval75.1%
Terminal-Bench 2.1 (Vals)84.6%
SWE-bench (Vals)97.0%
GPQA Diamond (Vals)93.4%
Vals Index67.2
EnigmaEval37.1%
HiL-Bench57.0%
EQ-Bench 41385
ARC-AGI-197.0%
ARC-AGI-285.4%
ARC-AGI-32.2%

Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.