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.

Specificationo4 MiniOpenAIClaude Opus 5Anthropic
ReleasedApr 16, 2025Jul 24, 2026
Context window200K1M
Max output100K128K
Input / 1M$1.1$5
Output / 1M$4.4$25
Cost per task$0.41$2.25
Arena Elo—1,393
Serving providers15
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

  • o4 Mini
  • Claude Opus 5

  • Intelligence Index

    o4 Mini
    26.1
    Claude Opus 5
    63.1

  • Coding Index

    o4 Mininot measured
    Claude Opus 5
    78.0

  • Agentic Index

    o4 Mininot measured
    Claude Opus 5
    59.2

  • Output speed

    o4 Mini
    108 t/s
    Claude Opus 5
    87 t/s

  • Context window

    o4 Mini
    200K
    Claude Opus 5
    1M

  • Latency · lower is better

    o4 Mini
    5.14s
    Claude Opus 5
    1.12s

  • Blended price / 1M · lower is better

    o4 Mini
    $1.93
    Claude Opus 5
    $10

  • Cost per task · lower is better

    o4 Mini
    $0.41
    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
Metrico4 MiniClaude Opus 5
Intelligence Index26.163.1
Coding Index—78.0
Agentic Index—59.2
Output speed108 t/s87 t/s
Context window200K1M
Latency · lower is better5.14s1.12s
Blended price / 1M · lower is better$1.93$10
Cost per task · lower is better$0.41$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • o4 Mini
  • Claude Opus 5

  • GPQA Diamond

    o4 Mini
    78.4%
    Claude Opus 5
    93.2%

  • Humanity's Last Exam

    o4 Mini
    16.5%
    Claude Opus 5
    54.9%

  • SciCode

    o4 Mini
    46.5%
    Claude Opus 5
    55.7%

  • τ²-bench

    o4 Mini
    55.6%
    Claude Opus 5
    42.1%

  • Terminal-Bench Hard

    o4 Mini
    15.2%
    Claude Opus 5not measured

  • LiveCodeBench

    o4 Mini
    85.9%
    Claude Opus 5not measured

  • AA-LCR long context

    o4 Mini
    60.0%
    Claude Opus 5
    75.7%

  • AIME 2025

    o4 Mini
    90.7%
    Claude Opus 5not measured

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

View as table
Evaluationo4 MiniClaude Opus 5
GPQA Diamond78.4%93.2%
Humanity's Last Exam16.5%54.9%
SciCode46.5%55.7%
τ²-bench55.6%42.1%
Terminal-Bench Hard15.2%—
LiveCodeBench85.9%—
AA-LCR long context60.0%75.7%
AIME 202590.7%—

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