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 Opus 4.7 vs Claude Opus 4.8

Verdict

Claude Opus 4.8 leads 12–4 across 16 shared benchmarks.

Claude Opus 4.7 4 · Claude Opus 4.8 12 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Claude Opus 4.7Claude Opus 4.8
  • 33.9
    APEX-Agents
    42.5
  • 93.5
    ARC-AGI
    92.5
  • 75.8
    ARC-AGI-2
    72.08
  • 30
    Chess Puzzles
    34
  • 12
    CritPt
    20.86
  • 64.8
    CursorBench
    63.8
  • 31.71
    FrontierMath-Tier-4-v2-Private
    56.1
  • 70.18
    FrontierMath-Tiers-1-3-v2-Private
    80
  • 43.81
    GBAEval
    70.9
  • 86.87
    GPQA diamond
    88.05
  • 18.2
    OSWorld 2.0
    20.6
  • 97.8
    OTIS Mock AIME 2024-2025
    98.33
  • 28.56
    PostTrainBench
    34.08
  • 55.48
    SimpleBench
    57.76
  • 50.6
    SimpleQA Verified
    39.5
  • 76.4
    WeirdML
    82.89
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 Opus 4.797.8
  • Claude Opus 4.898.33