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 Flash vs Gemini 3 Pro

Verdict

Gemini 3 Pro leads 9–6 across 15 shared benchmarks.

Gemini 3 Flash 6 · Gemini 3 Pro 9 · higher isn't always better — see caveats.

Per-benchmark head-to-head

Gemini 3 FlashGemini 3 Pro
  • 24
    APEX-Agents
    18.4
  • 21.5
    ARC-AGI
    75
  • 33.61
    ARC-AGI-2
    31.11
  • 48.1
    Balrog
    58.1
  • 38
    Chess Puzzles
    31
  • 88
    GeoBench
    84
  • 77.61
    GPQA diamond
    90.15
  • 9.8
    GSO-Bench
    18.6
  • 92.77
    OTIS Mock AIME 2024-2025
    91.38
  • 53.32
    SimpleBench
    71.68
  • 67.4
    SimpleQA Verified
    72.9
  • 75.41
    SWE-Bench verified
    72.93
  • 64.3
    Terminal Bench
    69.4
  • 58.9
    VPCT
    86.5
  • 61.6
    WeirdML
    69.93
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 Flash92.77
  • Gemini 3 Pro91.38