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.

← Integrity Index

FrontierMath-Tier-4-2025-07-01-Private

FrontierMath-Tier-4-2025-07-01-Private

Rank #13 of 51 · 27 models scored · top score 31.25

AINTEGRITY 93 / 100

unknown

Strongest on contamination resistance, weakest on discrimination.

consistent harnessheld-out test set

Headroom to ceiling

68.8 headroomtop 31.3ceiling 100

Score distribution

median 3.5Gemini 3 Pro 31.3016.533

State of the art over time

0173331.320242026

Integrity breakdown

Weighted composite. Each signal is computed from the source-backed leaderboard below.

Discrimination30%78

Does it still separate models?

Saturation headroom30%99

How far from ceiling / clustered at the top?

Contamination resistance15%100

Public vs held-out; training-leak risk.

Harness comparability15%100

Apples-to-apples, or mixed / vendor-optimized?

Freshness10%100

How old is the benchmark?

Leaderboard

#ModelScoreEvidenceMeasured
1Gemini 3 Pro31.25reproduced2025-11-18
2Muse Spark24.33reproduced2026-04-08
3GLM-5.120.83reproduced2026-04-07
4GPT-5.120.83reproduced2025-11-13
5Qwen 3.6 Plus13.89reproduced2026-04-01
6Claude Sonnet 4.613.83reproduced2026-02-17
7Kimi K2.57reproduced2026-02-02
8Qwen 3.6 Max (Preview)6.94reproduced2026-04-20
9Claude Opus 46.94reproduced2025-05-22
10Gemini 2.5 Flash (Jun 2025)6.94reproduced2025-06-17
11GLM-4.63.55reproduced2025-09-30
12DeepSeek-V3.23.5reproduced2025-12-01
13GLM-53.5reproduced2026-02-11
14Grok 43.47reproduced2025-07-09
15Claude Haiku 4.53.47reproduced2025-10-15
16Qwen 3.5 Plus (hosted 397B-A17B)3.47reproduced2026-02-16
17o33.47reproduced2024-12-20
18Grok 30reproduced2025-02-17
19Kimi K2 Thinking0reproduced2025-11-06
20Claude 3.5 Sonnet0reproduced2024-06-20
21Claude 3.5 Sonnet (October 2024)0reproduced2024-10-22
22Qwen 3.5 Flash (hosted 35B-A3B)0reproduced2026-02-25
23Qwen 3.6 Flash0reproduced2026-04-26
24Qwen3-235B-A22B-Thinking (Jul 2025)0reproduced2025-07-25
25Claude Sonnet 40reproduced2025-05-22
26GLM-4.70reproduced2025-12-22
27GPT-4.10reproduced2025-04-14

At a glance

Models scored
27
Top model
Gemini 3 Pro
Top score
31.25
Score spread
31.25
Headroom
68.75

Benchmark context

Task type
math
Metric
normalized performance in Epoch CSV, rendered as accuracy (%)
Score ceiling
100
Human baseline
unknown
Test-set privacy
held-out
Construction
unknown
Key assumptions
unknown
Limitations
unknown
Contamination history
unknown
Harness notes
Epoch CSV reports model, model version, performance, optimized flag, score source, benchmark release date, and model date.
Authors
unknown
Released
unknown
License
CC BY 4.0
Source
epoch.ai/data/eci_benchmarks.csv

Provenance

27 source records · scores from epoch_benchmarks. Every fact on this page is evidence-linked.

Follow the record

Integrity neighbors

  • FrontierMath-Tiers-1-3-v2-PrivateA · 98
  • GBAEvalA · 98
  • GDPvalA · 98
  • CritPtA · 92
  • The Agent CompanyA · 92
  • GSO-BenchA · 90

Top models here

  • Gemini 3 Pro31.25
  • Muse Spark24.33
  • GLM-5.120.83
  • GPT-5.120.83
  • Qwen 3.6 Plus13.89

Head-to-head

  • Gemini 3 Pro vs Muse Spark
  • Gemini 3 Pro vs GLM-5.1
  • Muse Spark vs GLM-5.1