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.

SpecificationClaude Opus 5AnthropicClaude Fable 5Anthropic
ReleasedJul 24, 2026Jun 9, 2026
Context window1M1M
Max output128K128K
Input / 1M$5$10
Output / 1M$25$50
Cost per task$2.25$4.50
Arena Elo1,3931,397
Serving providers54
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

  • Claude Opus 5
  • Claude Fable 5

  • Intelligence Index

    Claude Opus 5
    63.1
    Claude Fable 5
    62.1

  • Coding Index

    Claude Opus 5
    78.0
    Claude Fable 5
    76.5

  • Agentic Index

    Claude Opus 5
    59.2
    Claude Fable 5
    56.6

  • Output speed

    Claude Opus 5
    87 t/s
    Claude Fable 5
    54 t/s

  • Context window

    Claude Opus 5
    1M
    Claude Fable 5
    1M

  • Latency · lower is better

    Claude Opus 5
    1.12s
    Claude Fable 5
    3.84s

  • Blended price / 1M · lower is better

    Claude Opus 5
    $10
    Claude Fable 5
    $20

  • Cost per task · lower is better

    Claude Opus 5
    $2.25
    Claude Fable 5
    $4.50

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
MetricClaude Opus 5Claude Fable 5
Intelligence Index63.162.1
Coding Index78.076.5
Agentic Index59.256.6
Output speed87 t/s54 t/s
Context window1M1M
Latency · lower is better1.12s3.84s
Blended price / 1M · lower is better$10$20
Cost per task · lower is better$2.25$4.50

Evaluation scores

Percentage correct on a common 0–100% scale

  • Claude Opus 5
  • Claude Fable 5

  • GPQA Diamond

    Claude Opus 5
    93.2%
    Claude Fable 5
    92.6%

  • Humanity's Last Exam

    Claude Opus 5
    54.9%
    Claude Fable 5
    55.5%

  • SciCode

    Claude Opus 5
    55.7%
    Claude Fable 5
    60.2%

  • τ²-bench

    Claude Opus 5
    42.1%
    Claude Fable 5
    98.5%

  • Terminal-Bench Hard

    Claude Opus 5not measured
    Claude Fable 5
    62.9%

  • AA-LCR long context

    Claude Opus 5
    75.7%
    Claude Fable 5
    76.7%

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

View as table
EvaluationClaude Opus 5Claude Fable 5
GPQA Diamond93.2%92.6%
Humanity's Last Exam54.9%55.5%
SciCode55.7%60.2%
τ²-bench42.1%98.5%
Terminal-Bench Hard—62.9%
LiveCodeBench——
AA-LCR long context75.7%76.7%
AIME 2025——

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