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

o3 vs o4-mini

Verdict

o3 leads 15–3 across 18 shared benchmarks.

o3 15 · o4-mini 3 · higher isn't always better — see caveats.

Per-benchmark head-to-head

o3o4-mini
  • 81.3
    Aider polyglot
    72
  • 60.8
    ARC-AGI
    58.7
  • 6.53
    ARC-AGI-2
    6.11
  • 74
    CadEval
    62
  • 1.4
    CritPt
    0.6
  • 88.9
    Fiction.LiveBench
    77.8
  • 30.8
    GDPval
    25.3
  • 74
    GeoBench
    64
  • 75.76
    GPQA diamond
    72.81
  • 8.8
    GSO-Bench
    3.6
  • 16.3
    HLE
    13.95
  • 83.9
    Lech Mazur Writing
    75
  • 97.77
    MATH level 5
    97.83
  • 83.87
    OTIS Mock AIME 2024-2025
    81.65
  • 43.72
    SimpleBench
    26.44
  • 53
    SimpleQA Verified
    23.9
  • 28
    VPCT
    36.25
  • 52.42
    WeirdML
    52.56
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

  • o397.77
  • o4-mini97.83