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.

SpecificationMercury 2InceptionClaude Opus 5Anthropic
ReleasedMar 4, 2026Jul 24, 2026
Context window128K1M
Max output50K128K
Input / 1M$0.25$5
Output / 1M$0.75$25
Cost per task$0.07$2.25
Arena Elo1,0381,393
Serving providers15
ParametersUndisclosedUndisclosed
LicenceProprietaryProprietary
Capabilities
  • Reasoning
  • Tool use
  • 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

  • Mercury 2
  • Claude Opus 5

  • Intelligence Index

    Mercury 2
    21.9
    Claude Opus 5
    63.1

  • Coding Index

    Mercury 2
    31.1
    Claude Opus 5
    78.0

  • Agentic Index

    Mercury 2
    9.5
    Claude Opus 5
    59.2

  • Output speed

    Mercury 2
    228 t/s
    Claude Opus 5
    87 t/s

  • Context window

    Mercury 2
    128K
    Claude Opus 5
    1M

  • Latency · lower is better

    Mercury 2
    572ms
    Claude Opus 5
    1.12s

  • Blended price / 1M · lower is better

    Mercury 2
    $0.375
    Claude Opus 5
    $10

  • Cost per task · lower is better

    Mercury 2
    $0.07
    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
MetricMercury 2Claude Opus 5
Intelligence Index21.963.1
Coding Index31.178.0
Agentic Index9.559.2
Output speed228 t/s87 t/s
Context window128K1M
Latency · lower is better572ms1.12s
Blended price / 1M · lower is better$0.375$10
Cost per task · lower is better$0.07$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • Mercury 2
  • Claude Opus 5

  • GPQA Diamond

    Mercury 2
    77.0%
    Claude Opus 5
    93.2%

  • Humanity's Last Exam

    Mercury 2
    17.1%
    Claude Opus 5
    54.9%

  • SciCode

    Mercury 2
    38.7%
    Claude Opus 5
    55.7%

  • τ²-bench

    Mercury 2
    70.8%
    Claude Opus 5
    42.1%

  • Terminal-Bench Hard

    Mercury 2
    26.5%
    Claude Opus 5not measured

  • AA-LCR long context

    Mercury 2
    40.7%
    Claude Opus 5
    75.7%

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

View as table
EvaluationMercury 2Claude Opus 5
GPQA Diamond77.0%93.2%
Humanity's Last Exam17.1%54.9%
SciCode38.7%55.7%
τ²-bench70.8%42.1%
Terminal-Bench Hard26.5%—
LiveCodeBench——
AA-LCR long context40.7%75.7%
AIME 2025——

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