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 4.6 vs Claude Opus 4.8

Verdict

Claude Opus 4.8 leads 11–4 across 15 shared benchmarks.

Claude Opus 4.6 4 · Claude Opus 4.8 11 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Claude Opus 4.6Claude Opus 4.8
  • 31.7
    APEX-Agents
    42.5
  • 94
    ARC-AGI
    92.5
  • 69.17
    ARC-AGI-2
    72.08
  • 17
    Chess Puzzles
    34
  • 55.31
    DeepResearch Bench
    50.23
  • 26.83
    FrontierMath-Tier-4-v2-Private
    56.1
  • 65.96
    FrontierMath-Tiers-1-3-v2-Private
    80
  • 44.12
    GBAEval
    70.9
  • 87.37
    GPQA diamond
    88.05
  • 94.44
    OTIS Mock AIME 2024-2025
    98.33
  • 24.82
    PostTrainBench
    34.08
  • 4.17
    Remote Labor Index
    8.33
  • 61.12
    SimpleBench
    57.76
  • 46.49
    SimpleQA Verified
    39.5
  • 77.95
    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 4.694.44
  • Claude Opus 4.898.33