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.

SpecificationKimi K2.7 CodeMoonshot AIClaude Opus 5Anthropic
ReleasedJun 12, 2026Jul 24, 2026
Context window262K1M
Max output262K128K
Input / 1M$0.7$5
Output / 1M$3.5$25
Cost per task$0.32$2.25
Arena Elo1,2981,393
Serving providers145
Parameters1026.9BUndisclosed
LicenceotherProprietary
Capabilities
  • Reasoning
  • 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

  • Kimi K2.7 Code
  • Claude Opus 5
  • Intelligence Index

    Kimi K2.7 Code
    43.0
    Claude Opus 5
    63.1
  • Coding Index

    Kimi K2.7 Code
    60.8
    Claude Opus 5
    78.0
  • Agentic Index

    Kimi K2.7 Code
    30.3
    Claude Opus 5
    59.2
  • Output speed

    Kimi K2.7 Code
    140 t/s
    Claude Opus 5
    87 t/s
  • Context window

    Kimi K2.7 Code
    262K
    Claude Opus 5
    1M
  • Latency · lower is better

    Kimi K2.7 Code
    621ms
    Claude Opus 5
    1.12s
  • Blended price / 1M · lower is better

    Kimi K2.7 Code
    $1.4
    Claude Opus 5
    $10
  • Cost per task · lower is better

    Kimi K2.7 Code
    $0.32
    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
MetricKimi K2.7 CodeClaude Opus 5
Intelligence Index43.063.1
Coding Index60.878.0
Agentic Index30.359.2
Output speed140 t/s87 t/s
Context window262K1M
Latency · lower is better621ms1.12s
Blended price / 1M · lower is better$1.4$10
Cost per task · lower is better$0.32$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • Kimi K2.7 Code
  • Claude Opus 5
  • GPQA Diamond

    Kimi K2.7 Code
    89.6%
    Claude Opus 5
    93.2%
  • Humanity's Last Exam

    Kimi K2.7 Code
    35.0%
    Claude Opus 5
    54.9%
  • SciCode

    Kimi K2.7 Code
    47.5%
    Claude Opus 5
    55.7%
  • τ²-bench

    Kimi K2.7 Code
    90.1%
    Claude Opus 5
    42.1%
  • Terminal-Bench Hard

    Kimi K2.7 Code
    44.7%
    Claude Opus 5not measured
  • AA-LCR long context

    Kimi K2.7 Code
    75.0%
    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
EvaluationKimi K2.7 CodeClaude Opus 5
GPQA Diamond89.6%93.2%
Humanity's Last Exam35.0%54.9%
SciCode47.5%55.7%
τ²-bench90.1%42.1%
Terminal-Bench Hard44.7%
LiveCodeBench
AA-LCR long context75.0%75.7%
AIME 2025

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