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

Qwen2-72B

Qwen·released 2024-06-07

independently reproduced

Scored on 5 benchmarks. 40% of its scores are independently reproduced — the rest are self-reported or unverified.

5
Benchmarks scored
0
SOTA leads
76.53
Best score
40%
Reproduced
2 of 5

Benchmark fingerprint

Every score, ranked, colored by evidence status.

  • MMLU
    76.53
  • MATH level 5
    39.07
  • GPQA diamond
    21.04
  • WeirdML
    11.3
  • The Agent Company
    1.1

All benchmark scores

BenchmarkScoreEvidence statusMeasured
MMLU76.53accuracy (%)unverified· optimized2024-06-07
MATH level 539.07accuracy (%)reproduced· optimized2024-06-07
GPQA diamond21.04accuracy (%)reproduced· optimized2024-06-07
WeirdML11.3accuracy (%)unverified2024-06-07
The Agent Company1.1accuracy (%)unverified2024-06-07
5 source records · epoch_benchmarks. Every score is evidence-linked.

Follow the record

More from Qwen

  • Qwen3-Max97.13
  • Qwen3.7-Max94.99
  • Qwen2.5-72B92.67
  • Qwen 3.6 Max (Preview)91.1
  • Qwen2.5-Coder-32B91.1
  • Qwen 3.6 Plus90.55