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 GPT-5

Verdict

Claude Opus 4.6 leads 15–2 across 17 shared benchmarks.

Claude Opus 4.6 15 · GPT-5 2 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Claude Opus 4.6GPT-5
  • 31.7
    APEX-Agents
    18.3
  • 94
    ARC-AGI
    65.7
  • 69.17
    ARC-AGI-2
    9.86
  • 17
    Chess Puzzles
    37
  • 55.31
    DeepResearch Bench
    55.13
  • 26.83
    FrontierMath-Tier-4-v2-Private
    21.95
  • 65.96
    FrontierMath-Tiers-1-3-v2-Private
    55.44
  • 87.37
    GPQA diamond
    81.57
  • 41.2
    GSO-Bench
    6.9
  • 31.13
    HLE
    21.55
  • 94.44
    OTIS Mock AIME 2024-2025
    91.38
  • 4.17
    Remote Labor Index
    1.67
  • 61.12
    SimpleBench
    48.04
  • 46.49
    SimpleQA Verified
    50.6
  • 78.72
    SWE-Bench verified
    73.55
  • 79.8
    Terminal Bench
    49.6
  • 77.95
    WeirdML
    60.7
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
  • GPT-598.13