EBR-bench
Integrity rank #16 of 65 · 25 models scored · top score 76.19 · GPT-6 Astra
unknown
Strongest on discrimination, weakest on contamination resistance. Caveat: the public test set does not rule out training-data contamination.
Reported scores — protocols may differ. How we compare
| # | Model | Score | Evidence | Measured |
|---|---|---|---|---|
| 1 | GPT-6 Astra | 76.19 | unverifiedT2 | 2026-09-03 |
| 2 | Claude Opus 5.5 | 71.43 | unverifiedT2 | 2026-09-22 |
| 3 | Claude Fable 5.1 | 57.14 | unverifiedT2 | 2026-09-01 |
| 4 | GPT-6.1 Sol | 54.29 | unverifiedT2 | 2026-09-29 |
| 5 | GPT-6 Sol | 53.33 | unverifiedT2 | 2026-09-22 |
| 6 | Claude Opus 5 | 45.71 | unverifiedT2 | 2026-07-24 |
| 7 | GPT-5.6 Sol | 44.76 | unverifiedT2 | 2026-07-09 |
| 8 | Claude Haiku 5.5 | 43.81 | unverifiedT2 | 2026-10-07 |
| 9 | Claude Fable 5 | 39.52 | reproducedT1 | 2026-06-09 |
| 10 | GPT-5.5 | 34.29 | reproducedT1 | 2026-04-23 |
| 11 | Grok 4.6 | 30.48 | reproducedT1 | 2026-08-12 |
| 12 | Claude Opus 4.8 | 28.57 | reproducedT1 | 2026-05-28 |
| 13 | GPT-5.4 | 25.4 | reproducedT1 | 2026-03-05 |
| 14 | GPT-5.2 | 23.02 | reproducedT1 | 2025-12-11 |
| 15 | Claude Opus 4.7 | 19.05 | reproducedT1 | 2026-04-16 |
| 16 | Gemini 3.1 Pro | 14.29 | reproducedT1 | 2026-02-19 |
| 17 | Claude Opus 4.5 | 14.29 | reproducedT1 | 2025-11-24 |
| 18 | Claude Opus 4.6 | 12.7 | reproducedT1 | 2026-02-05 |
| 19 | GPT-5 | 12.7 | reproducedT1 | 2025-08-07 |
| 20 | GLM-5.2 | 9.52 | reproducedT1 | 2026-06-16 |
| 21 | Qwen3.7-Max | 9.52 | reproducedT1 | 2026-05-19 |
| 22 | Claude Opus 4.1 | 7.94 | reproducedT1 | 2025-08-05 |
| 23 | Gemini 3.5 Flash | 4.76 | reproducedT1 | 2026-05-19 |
| 24 | Kimi K2.6 | 2.38 | reproducedT1 | 2026-04-20 |
| 25 | Claude Sonnet 4.5 | 2.38 | reproducedT1 | 2025-09-29 |
Continue from this leaderboard into the tightest head-to-head reads.
Weighted composite. Each dimension opens to the source-backed sub-signals it is computed from.
Claims drawn from cited facts, not live model generation.
The recorded metadata does not specify what this benchmark measures.
1 cited fact
This benchmark has scores from 25 tracked models, with a recorded score range spanning 73.81.
2 cited facts
Its Benchmark Integrity Index score is 94, with grade A and rank 16 out of 65 tracked benchmarks.
4 cited facts
It is not flagged as saturated. The top recorded score is 76.19, with 23.81 points of headroom. The computed top-score cluster contains 1 models.
4 cited facts
The harness classification is consistent harness. Test-set privacy is recorded as unknown. The recorded contamination-window ratio is 0.36.
3 cited facts
EBR-bench: EBR-bench as reported in Epoch AI's Capabilities Index CSV.
GPT-6 Astra leads EBR-bench at 76.19. The full leaderboard above lists every recorded measurement, not just the headline number.
25 models have recorded scores on EBR-bench, spanning a score spread of 73.81.
tensor.news grades EBR-bench A for integrity (score 94/100), ranking #16 of 65 benchmarks we assess across discrimination, saturation, contamination resistance, harness comparability, and freshness.
No — EBR-bench still has headroom and continues to discriminate between models rather than bunching them at the ceiling.
Its test set is unknown, so contamination risk is on the table: a high score may partly reflect training-data overlap rather than capability.
Scores are reported under a consistent harness, so comparisons on EBR-bench are reasonably apples-to-apples.
Every EBR-bench measurement is source-backed and tagged reproduced, self-reported, or unverified; the underlying record cites epoch.ai/data/eci_benchmarks.csv.