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 3.1 Pro vs GPT-5.5

Verdict

GPT-5.5 leads 14–5 across 19 shared benchmarks.

Gemini 3.1 Pro 5 · GPT-5.5 14 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Gemini 3.1 ProGPT-5.5
  • 33.5
    APEX-Agents
    38.4
  • 98
    ARC-AGI
    95
  • 77.1
    ARC-AGI-2
    85
  • 55
    Chess Puzzles
    54
  • 16.9
    CL-bench Life
    22.2
  • 17.71
    CritPt
    27.14
  • 26
    ExploitBench
    41
  • 26.83
    FrontierMath-Tier-4-v2-Private
    72.5
  • 59.65
    FrontierMath-Tiers-1-3-v2-Private
    85.26
  • 0.85
    GBAEval
    53.22
  • 92.13
    GPQA diamond
    92
  • 22.55
    GSO-Bench
    40.2
  • 95.6
    OTIS Mock AIME 2024-2025
    100
  • 21.59
    PostTrainBench
    25.02
  • 75.52
    SimpleBench
    62.8
  • 77.3
    SimpleQA Verified
    63.1
  • 75.62
    SWE-Bench verified
    80.58
  • 80.2
    Terminal Bench
    84.7
  • 72.1
    WeirdML
    84.91
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 3.1 Pro98
  • GPT-5.5100

Gemini 3.1 Pro vs …

  • vs PowerMoE-3b
  • vs Claude Fable 5
  • vs Llama 3.1-405B
  • vs phi-3-small 7.4B
  • vs DeepSeek-V3

GPT-5.5 vs …

  • vs PowerMoE-3b
  • vs Claude Fable 5
  • vs Llama 3.1-405B
  • vs phi-3-small 7.4B
  • vs DeepSeek-V3