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

Gemini 3 Pro vs GPT-5.2

Verdict

GPT-5.2 leads 12–6 across 18 shared benchmarks.

Gemini 3 Pro 6 · GPT-5.2 12 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Gemini 3 ProGPT-5.2
  • 18.4
    APEX-Agents
    34.3
  • 75
    ARC-AGI
    86.2
  • 31.11
    ARC-AGI-2
    52.91
  • 31
    Chess Puzzles
    49
  • 15.8
    CL-bench
    18.2
  • 40.3
    GDPval
    49.7
  • 90.15
    GPQA diamond
    88.53
  • 18.6
    GSO-Bench
    27.4
  • 34.37
    HLE
    24.16
  • 91.38
    OTIS Mock AIME 2024-2025
    96.11
  • 18.12
    PostTrainBench
    21.38
  • 1.25
    Remote Labor Index
    2.5
  • 71.68
    SimpleBench
    34.96
  • 72.9
    SimpleQA Verified
    38.9
  • 72.93
    SWE-Bench verified
    73.76
  • 69.4
    Terminal Bench
    64.9
  • 86.5
    VPCT
    76
  • 69.93
    WeirdML
    72.2
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

  • Gemini 3 Pro91.38
  • GPT-5.296.11