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.

SpecificationLlama 4 ScoutMetaClaude Opus 5Anthropic
ReleasedApr 5, 2025Jul 24, 2026
Context window1.31M1M
Max output16K128K
Input / 1M$0.1$5
Output / 1M$0.3$25
Cost per task$0.01$2.25
Arena Elo9231,393
Serving providers45
Parameters108.6BUndisclosed
LicenceotherProprietary
Capabilities
  • Tool use
  • Vision
  • Open weights
  • 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

  • Llama 4 Scout
  • Claude Opus 5
  • Intelligence Index

    Llama 4 Scout
    10.3
    Claude Opus 5
    63.1
  • Coding Index

    Llama 4 Scout
    8.2
    Claude Opus 5
    78.0
  • Agentic Index

    Llama 4 Scout
    1.1
    Claude Opus 5
    59.2
  • Output speed

    Llama 4 Scout
    152 t/s
    Claude Opus 5
    87 t/s
  • Context window

    Llama 4 Scout
    1.31M
    Claude Opus 5
    1M
  • Latency · lower is better

    Llama 4 Scout
    265ms
    Claude Opus 5
    1.12s
  • Blended price / 1M · lower is better

    Llama 4 Scout
    $0.15
    Claude Opus 5
    $10
  • Cost per task · lower is better

    Llama 4 Scout
    $0.01
    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
MetricLlama 4 ScoutClaude Opus 5
Intelligence Index10.363.1
Coding Index8.278.0
Agentic Index1.159.2
Output speed152 t/s87 t/s
Context window1.31M1M
Latency · lower is better265ms1.12s
Blended price / 1M · lower is better$0.15$10
Cost per task · lower is better$0.01$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • Llama 4 Scout
  • Claude Opus 5
  • GPQA Diamond

    Llama 4 Scout
    58.7%
    Claude Opus 5
    93.2%
  • Humanity's Last Exam

    Llama 4 Scout
    3.8%
    Claude Opus 5
    54.9%
  • SciCode

    Llama 4 Scout
    17.0%
    Claude Opus 5
    55.7%
  • τ²-bench

    Llama 4 Scout
    15.5%
    Claude Opus 5
    42.1%
  • Terminal-Bench Hard

    Llama 4 Scout
    1.5%
    Claude Opus 5not measured
  • LiveCodeBench

    Llama 4 Scout
    29.9%
    Claude Opus 5not measured
  • AA-LCR long context

    Llama 4 Scout
    30.3%
    Claude Opus 5
    75.7%
  • AIME 2025

    Llama 4 Scout
    14.0%
    Claude Opus 5not measured

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

View as table
EvaluationLlama 4 ScoutClaude Opus 5
GPQA Diamond58.7%93.2%
Humanity's Last Exam3.8%54.9%
SciCode17.0%55.7%
τ²-bench15.5%42.1%
Terminal-Bench Hard1.5%
LiveCodeBench29.9%
AA-LCR long context30.3%75.7%
AIME 202514.0%

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