tensor.news
Integrity IndexModelsCompareLabsPapersMethodologyAbout
tensor.news

© 2026

Evidence

  • Integrity Index
  • Model Claim Ledger
  • Compare

Entities

  • Models
  • Benchmarks
  • Labs
  • Papers

Method

  • Methodology
  • About
  • Privacy

Machine

  • Feed (Atom)
  • llms.txt
  • llms-full.txt
  • Sitemap

tensor.news grades benchmarks and tags model claims by evidence status. Every fact carries a source record and provenance. Scores are framed as task performance under a disclosed harness, never as deployed capability.

← Compare

Claude Sonnet 4.6 vs Kimi K2.5

Verdict

Claude Sonnet 4.6 leads 12–3 across 15 shared benchmarks.

Claude Sonnet 4.6 12 · Kimi K2.5 3 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Claude Sonnet 4.6Kimi K2.5
  • 23.7
    APEX-Agents
    14.4
  • 86.5
    ARC-AGI
    65.33
  • 60.42
    ARC-AGI-2
    11.81
  • 13
    Chess Puzzles
    12
  • 49
    CursorBench
    31.9
  • 56.84
    FrontierMath-2025-02-28-Private
    48.95
  • 13.83
    FrontierMath-Tier-4-2025-07-01-Private
    7
  • 83.16
    GPQA diamond
    83.47
  • 72.1
    OSWorld
    63.3
  • 85.78
    OTIS Mock AIME 2024-2025
    92.19
  • 16.42
    PostTrainBench
    10.26
  • 29
    SimpleQA Verified
    33.9
  • 75.21
    SWE-Bench verified
    73.76
  • 53.4
    Terminal Bench
    43.2
  • 66.07
    WeirdML
    45.6
A higher number is not always a better model. Each score is task performance under a disclosed harness, and some of these benchmarks are saturated or scored under mixed harnesses. Follow any benchmark to its integrity grade before reading a win as decisive.

Follow the record

These models

  • Claude Sonnet 4.686.5
  • Kimi K2.592.19