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.

SpecificationHy3TencentClaude Opus 5Anthropic
ReleasedJul 6, 2026Jul 24, 2026
Context window262K1M
Max output128K128K
Input / 1M$0.132$5
Output / 1M$0.528$25
Cost per task$0.05$2.25
Arena Elo1,2311,393
Serving providers65
Parameters298.8BUndisclosed
Licenceapache-2.0Proprietary
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

  • Hy3
  • Claude Opus 5

  • Intelligence Index

    Hy3
    42.2
    Claude Opus 5
    63.1

  • Coding Index

    Hy3
    58.8
    Claude Opus 5
    78.0

  • Agentic Index

    Hy3
    31.4
    Claude Opus 5
    59.2

  • Output speed

    Hy3
    78 t/s
    Claude Opus 5
    87 t/s

  • Context window

    Hy3
    262K
    Claude Opus 5
    1M

  • Latency · lower is better

    Hy3
    434ms
    Claude Opus 5
    1.12s

  • Blended price / 1M · lower is better

    Hy3
    $0.231
    Claude Opus 5
    $10

  • Cost per task · lower is better

    Hy3
    $0.05
    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
MetricHy3Claude Opus 5
Intelligence Index42.263.1
Coding Index58.878.0
Agentic Index31.459.2
Output speed78 t/s87 t/s
Context window262K1M
Latency · lower is better434ms1.12s
Blended price / 1M · lower is better$0.231$10
Cost per task · lower is better$0.05$2.25

Evaluation scores

Percentage correct on a common 0–100% scale

  • Hy3
  • Claude Opus 5

  • GPQA Diamond

    Hy3
    89.7%
    Claude Opus 5
    93.2%

  • Humanity's Last Exam

    Hy3
    33.5%
    Claude Opus 5
    54.9%

  • SciCode

    Hy3
    47.6%
    Claude Opus 5
    55.7%

  • τ²-bench

    Hy3
    22.9%
    Claude Opus 5
    42.1%

  • AA-LCR long context

    Hy3
    74.7%
    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
EvaluationHy3Claude Opus 5
GPQA Diamond89.7%93.2%
Humanity's Last Exam33.5%54.9%
SciCode47.6%55.7%
τ²-bench22.9%42.1%
Terminal-Bench Hard——
LiveCodeBench——
AA-LCR long context74.7%75.7%
AIME 2025——

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