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.

SpecificationM2.1MiniMaxClaude Opus 5Anthropic
ReleasedDec 23, 2025Jul 24, 2026
Context window205K1M
Max output131K128K
Input / 1M$0.3$5
Output / 1M$1.2$25
Cost per task$0.11$2.25
Arena Elo1,2581,393
Serving providers25
Parameters228.7BUndisclosed
LicenceotherProprietary
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

  • M2.1
  • Claude Opus 5
  • Intelligence Index

    M2.1
    32.1
    Claude Opus 5
    63.1
  • Coding Index

    M2.1not measured
    Claude Opus 5
    78.0
  • Agentic Index

    M2.1not measured
    Claude Opus 5
    59.2
  • Output speed

    M2.1
    66 t/s
    Claude Opus 5
    87 t/s
  • Context window

    M2.1
    205K
    Claude Opus 5
    1M
  • Latency · lower is better

    M2.1
    668ms
    Claude Opus 5
    1.12s
  • Blended price / 1M · lower is better

    M2.1
    $0.525
    Claude Opus 5
    $10
  • Cost per task · lower is better

    M2.1
    $0.11
    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
MetricM2.1Claude Opus 5
Intelligence Index32.163.1
Coding Index78.0
Agentic Index59.2
Output speed66 t/s87 t/s
Context window205K1M
Latency · lower is better668ms1.12s
Blended price / 1M · lower is better$0.525$10
Cost per task · lower is better$0.11$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • M2.1
  • Claude Opus 5
  • GPQA Diamond

    M2.1
    83.0%
    Claude Opus 5
    93.2%
  • Humanity's Last Exam

    M2.1
    23.2%
    Claude Opus 5
    54.9%
  • SciCode

    M2.1
    40.7%
    Claude Opus 5
    55.7%
  • τ²-bench

    M2.1
    85.4%
    Claude Opus 5
    42.1%
  • Terminal-Bench Hard

    M2.1
    28.8%
    Claude Opus 5not measured
  • LiveCodeBench

    M2.1
    81.0%
    Claude Opus 5not measured
  • AA-LCR long context

    M2.1
    66.3%
    Claude Opus 5
    75.7%
  • AIME 2025

    M2.1
    82.7%
    Claude Opus 5not measured

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

View as table
EvaluationM2.1Claude Opus 5
GPQA Diamond83.0%93.2%
Humanity's Last Exam23.2%54.9%
SciCode40.7%55.7%
τ²-bench85.4%42.1%
Terminal-Bench Hard28.8%
LiveCodeBench81.0%
AA-LCR long context66.3%75.7%
AIME 202582.7%

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