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.

SpecificationGLM 4.6Z.aiClaude Opus 5Anthropic
ReleasedSep 30, 2025Jul 24, 2026
Context window205K1M
Max output131K128K
Input / 1M$0.5$5
Output / 1M$2$25
Cost per task$0.18$2.25
Arena Elo1,1971,393
Serving providers55
Parameters356.8BUndisclosed
LicencemitProprietary
Capabilities
  • Reasoning
  • Tool use
  • 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

  • GLM 4.6
  • Claude Opus 5
  • Intelligence Index

    GLM 4.6
    23.4
    Claude Opus 5
    63.1
  • Coding Index

    GLM 4.6not measured
    Claude Opus 5
    78.0
  • Agentic Index

    GLM 4.6
    18.6
    Claude Opus 5
    59.2
  • Output speed

    GLM 4.6
    56 t/s
    Claude Opus 5
    87 t/s
  • Context window

    GLM 4.6
    205K
    Claude Opus 5
    1M
  • Latency · lower is better

    GLM 4.6
    451ms
    Claude Opus 5
    1.12s
  • Blended price / 1M · lower is better

    GLM 4.6
    $0.875
    Claude Opus 5
    $10
  • Cost per task · lower is better

    GLM 4.6
    $0.18
    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
MetricGLM 4.6Claude Opus 5
Intelligence Index23.463.1
Coding Index78.0
Agentic Index18.659.2
Output speed56 t/s87 t/s
Context window205K1M
Latency · lower is better451ms1.12s
Blended price / 1M · lower is better$0.875$10
Cost per task · lower is better$0.18$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • GLM 4.6
  • Claude Opus 5
  • GPQA Diamond

    GLM 4.6
    63.2%
    Claude Opus 5
    93.2%
  • Humanity's Last Exam

    GLM 4.6
    5.5%
    Claude Opus 5
    54.9%
  • SciCode

    GLM 4.6
    33.1%
    Claude Opus 5
    55.7%
  • τ²-bench

    GLM 4.6
    76.9%
    Claude Opus 5
    42.1%
  • Terminal-Bench Hard

    GLM 4.6
    28.8%
    Claude Opus 5not measured
  • LiveCodeBench

    GLM 4.6
    56.1%
    Claude Opus 5not measured
  • AA-LCR long context

    GLM 4.6
    28.3%
    Claude Opus 5
    75.7%
  • AIME 2025

    GLM 4.6
    44.3%
    Claude Opus 5not measured

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

View as table
EvaluationGLM 4.6Claude Opus 5
GPQA Diamond63.2%93.2%
Humanity's Last Exam5.5%54.9%
SciCode33.1%55.7%
τ²-bench76.9%42.1%
Terminal-Bench Hard28.8%
LiveCodeBench56.1%
AA-LCR long context28.3%75.7%
AIME 202544.3%

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