Evaluation sources Official evaluation
EBR-bench
Epoch AI / Professional work · Learns new rules and uses memory in long tasks
Across NexGateHub AI News Reference only
Evidence budgetNot scored
Upstream data time10/14 01:16
Last successful sync10/18 14:05
What it measures and how
Public aggregates mix memory-compression configurations. Reference only until fixed-configuration comparisons are available.
How this evidence is used
Reference only; excluded from overall and category scores.
Evaluation results
Fixed rules select one representative configuration per public model. Anonymous test models are omitted. Original ranks are retained, with up to 30 models per evaluation.
| Source rank | Source model name | Raw result | Representative configuration |
|---|---|---|---|
| 1 | gpt-6-astra_maxOpenAI | 76.2% | Max reasoning |
| 2 | claude-opus-5-5_maxAnthropic | 71.4% | Max reasoning |
| 3 | gemini-4-argon_highgoogle | 64.8% | High reasoning |
| 4 | claude-fable-5-1_maxAnthropic | 57.1% | Max reasoning |
| 5 | gpt-6.1-sol_maxopenai | 54.3% | Max reasoning |
| 6 | gpt-6-sol_maxOpenAI | 53.3% | Max reasoning |
| 7 | claude-opus-5_maxAnthropic | 45.7% | Max reasoning |
| 8 | gpt-5.6-sol_maxOpenAI | 44.8% | Max reasoning |
| 9 | claude-haiku-5-5_max— | 43.8% | Max reasoning |
| 10 | claude-fable-5_maxAnthropic | 39.5% | Max reasoning |
| 11 | gpt-5.5_xhighOpenAI | 34.3% | xHigh reasoning |
| 12 | muse-wonder-1.0_max— | 31.4% | Max reasoning |
| 13 | grok-4.6_xhighxAI | 30.5% | xHigh reasoning |
| 14 | claude-opus-4-8_maxAnthropic | 28.6% | Max reasoning |
| 15 | gpt-5.4-2026-03-05_xhighOpenAI | 25.4% | xHigh reasoning |
| 16 | gpt-5.2-2025-12-11_xhighOpenAI | 23.0% | xHigh reasoning |
| 17 | claude-opus-4-7_maxAnthropic | 19.0% | Max reasoning |
| 18 | claude-opus-4-5-20251101_128KAnthropic | 14.3% | Source default · 128k |
| 18 | gemini-3.1-pro-previewGoogle | 14.3% | Source default |
| 20 | gpt-5-2025-08-07_highOpenAI | 12.7% | High reasoning |
| 20 | claude-opus-4-6_maxAnthropic | 12.7% | Max reasoning |
| 22 | qwen3.7-maxAlibaba | 9.5% | Source default |
| 22 | glm-5.2_maxZ.ai | 9.5% | Max reasoning |
| 24 | claude-opus-4-1-20250805Anthropic | 7.9% | Source default |
| 25 | gemini-3.5-flash_highGoogle | 4.8% | High reasoning |
| 26 | kimi-k2.6Moonshot AI | 2.4% | Source default |
| 26 | claude-sonnet-4-5-20250929Anthropic | 2.4% | Source default |
Evaluation limits & attribution
Reference only; excluded from overall and category scores.
Data license: CC BY 4.0 · Epoch AI's own evaluation data.
Results are published by Epoch AI. Original scores and the NexGateHub consensus index use different scales and cannot be added directly.