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.

← Models

Gemma 3 27B

Google DeepMind·released 2025-03-12

independently reproduced

Scored on 8 benchmarks. 38% of its scores are independently reproduced — the rest are self-reported or unverified.

8
Benchmarks scored
0
SOTA leads
79.9
Best score
38%
Reproduced
3 of 8

Benchmark fingerprint

Every score, ranked, colored by evidence status.

  • Lech Mazur Writing
    79.9
  • MATH level 5
    74.04
  • GeoBench
    52
  • Fiction.LiveBench
    33.3
  • GPQA diamond
    31.82
  • OTIS Mock AIME 2024-2025
    19.64
  • Aider polyglot
    4.9
  • CritPt
    0

All benchmark scores

BenchmarkScoreEvidence statusMeasured
Lech Mazur Writing79.9accuracy (%)unverified2025-03-12
MATH level 574.04accuracy (%)reproduced· optimized2025-03-12
GeoBench52accuracy (%)unverified2025-03-12
Fiction.LiveBench33.3accuracy (%)unverified2025-03-12
GPQA diamond31.82accuracy (%)reproduced· optimized2025-03-12
OTIS Mock AIME 2024-202519.64accuracy (%)reproduced· optimized2025-03-12
Aider polyglot4.9accuracy (%)unverified· optimized2025-03-12
CritPt0accuracy (%)unverified2025-03-12
8 source records · epoch_benchmarks. Every score is evidence-linked.

Follow the record

More from Google DeepMind

  • Gemini 3.1 Pro98
  • Gemini 2.5 Pro (May 2025)95.9
  • Gemini 2.5 Pro (Mar 2025)95.56
  • Gemini 3.5 Flash95.55
  • Gemini 3 Flash92.77
  • Gemini 2.5 Pro (Jun 2025)91.7