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

GPT-5 vs Grok 4

Verdict

GPT-5 leads 12–5 across 17 shared benchmarks.

GPT-5 12 · Grok 4 5 · higher isn't always better — see caveats.

Per-benchmark head-to-head

GPT-5Grok 4
  • 88
    Aider polyglot
    79.6
  • 18.3
    APEX-Agents
    15.2
  • 65.7
    ARC-AGI
    66.67
  • 9.86
    ARC-AGI-2
    15.97
  • 32.8
    Balrog
    43.6
  • 37
    Chess Puzzles
    28
  • 55.13
    DeepResearch Bench
    47.9
  • 97.2
    Fiction.LiveBench
    94.4
  • 34.8
    GDPval
    21.1
  • 81
    GeoBench
    45
  • 81.57
    GPQA diamond
    82.67
  • 87.23
    Lech Mazur Writing
    80.68
  • 91.38
    OTIS Mock AIME 2024-2025
    83.98
  • 48.04
    SimpleBench
    52.6
  • 50.6
    SimpleQA Verified
    47.9
  • 49.6
    Terminal Bench
    27.2
  • 60.7
    WeirdML
    45.73
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

  • GPT-598.13
  • Grok 494.4