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 MaverickMetaClaude Opus 5Anthropic
ReleasedApr 5, 2025Jul 24, 2026
Context window1.05M1M
Max output16K128K
Input / 1M$0.2$5
Output / 1M$0.8$25
Cost per task$0.03$2.25
Arena Elo9551,393
Serving providers55
Parameters401.6BUndisclosed
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 Maverick
  • Claude Opus 5
  • Intelligence Index

    Llama 4 Maverick
    14.5
    Claude Opus 5
    63.1
  • Coding Index

    Llama 4 Maverick
    16.3
    Claude Opus 5
    78.0
  • Agentic Index

    Llama 4 Maverick
    1.2
    Claude Opus 5
    59.2
  • Output speed

    Llama 4 Maverick
    72 t/s
    Claude Opus 5
    87 t/s
  • Context window

    Llama 4 Maverick
    1.05M
    Claude Opus 5
    1M
  • Latency · lower is better

    Llama 4 Maverick
    324ms
    Claude Opus 5
    1.12s
  • Blended price / 1M · lower is better

    Llama 4 Maverick
    $0.35
    Claude Opus 5
    $10
  • Cost per task · lower is better

    Llama 4 Maverick
    $0.03
    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 MaverickClaude Opus 5
Intelligence Index14.563.1
Coding Index16.378.0
Agentic Index1.259.2
Output speed72 t/s87 t/s
Context window1.05M1M
Latency · lower is better324ms1.12s
Blended price / 1M · lower is better$0.35$10
Cost per task · lower is better$0.03$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • Llama 4 Maverick
  • Claude Opus 5
  • GPQA Diamond

    Llama 4 Maverick
    67.1%
    Claude Opus 5
    93.2%
  • Humanity's Last Exam

    Llama 4 Maverick
    4.9%
    Claude Opus 5
    54.9%
  • SciCode

    Llama 4 Maverick
    33.1%
    Claude Opus 5
    55.7%
  • τ²-bench

    Llama 4 Maverick
    17.8%
    Claude Opus 5
    42.1%
  • Terminal-Bench Hard

    Llama 4 Maverick
    6.8%
    Claude Opus 5not measured
  • LiveCodeBench

    Llama 4 Maverick
    39.7%
    Claude Opus 5not measured
  • AA-LCR long context

    Llama 4 Maverick
    50.0%
    Claude Opus 5
    75.7%
  • AIME 2025

    Llama 4 Maverick
    19.3%
    Claude Opus 5not measured

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

View as table
EvaluationLlama 4 MaverickClaude Opus 5
GPQA Diamond67.1%93.2%
Humanity's Last Exam4.9%54.9%
SciCode33.1%55.7%
τ²-bench17.8%42.1%
Terminal-Bench Hard6.8%
LiveCodeBench39.7%
AA-LCR long context50.0%75.7%
AIME 202519.3%

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