Skip to content
llmwaves

Compare

Four models, one page

Pick the models you are actually choosing between. Every row is scaled within itself, so a bar means something next to its neighbours and nothing across rows.

SpecificationGPT-5OpenAIClaude Opus 5Anthropic
ReleasedAug 7, 2025Jul 24, 2026
Context window400K1M
Max output128K128K
Input / 1M$1.25$5
Output / 1M$10$25
Cost per task$0.86$2.25
Arena Elo1,2621,393
Serving providers25
ParametersUndisclosedUndisclosed
LicenceProprietaryProprietary
Capabilities
  • Reasoning
  • Tool use
  • Vision
  • Reasoning
  • Tool use
  • Vision

Metrics side by side

Each row is scaled to the largest value in that row — bars compare within a row, never across rows

  • GPT-5
  • Claude Opus 5

  • Intelligence Index

    GPT-5
    35.3
    Claude Opus 5
    63.1

  • Coding Index

    GPT-5
    37.8
    Claude Opus 5
    78.0

  • Agentic Index

    GPT-5
    26.5
    Claude Opus 5
    59.2

  • Output speed

    GPT-5
    79 t/s
    Claude Opus 5
    87 t/s

  • Context window

    GPT-5
    400K
    Claude Opus 5
    1M

  • Latency · lower is better

    GPT-5
    5.3s
    Claude Opus 5
    1.12s

  • Blended price / 1M · lower is better

    GPT-5
    $3.44
    Claude Opus 5
    $10

  • Cost per task · lower is better

    GPT-5
    $0.86
    Claude Opus 5
    $2.25

Rows marked “lower is better” still draw a longer bar for a larger number — read the value, not just the length. Arena Elo is in the specification table above instead: it has no meaningful zero, so a bar would flatten the gaps.

View as table
MetricGPT-5Claude Opus 5
Intelligence Index35.363.1
Coding Index37.878.0
Agentic Index26.559.2
Output speed79 t/s87 t/s
Context window400K1M
Latency · lower is better5.3s1.12s
Blended price / 1M · lower is better$3.44$10
Cost per task · lower is better$0.86$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • GPT-5
  • Claude Opus 5

  • GPQA Diamond

    GPT-5
    85.4%
    Claude Opus 5
    93.2%

  • Humanity's Last Exam

    GPT-5
    28.5%
    Claude Opus 5
    54.9%

  • SciCode

    GPT-5
    42.9%
    Claude Opus 5
    55.7%

  • τ²-bench

    GPT-5
    84.8%
    Claude Opus 5
    42.1%

  • Terminal-Bench Hard

    GPT-5
    32.6%
    Claude Opus 5not measured

  • LiveCodeBench

    GPT-5
    84.6%
    Claude Opus 5not measured

  • AA-LCR long context

    GPT-5
    76.3%
    Claude Opus 5
    75.7%

  • AIME 2025

    GPT-5
    94.3%
    Claude Opus 5not measured

A missing bar means that evaluation was not run for that model — it is not a zero.

View as table
EvaluationGPT-5Claude Opus 5
GPQA Diamond85.4%93.2%
Humanity's Last Exam28.5%54.9%
SciCode42.9%55.7%
τ²-bench84.8%42.1%
Terminal-Bench Hard32.6%—
LiveCodeBench84.6%—
AA-LCR long context76.3%75.7%
AIME 202594.3%—

Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).