BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

EBR-bench leaderboard

As of 6 Nov 2026, the top score on EBR-bench is 78.1% by GPT-6.1 Terra (OpenAI), ahead of GPT-6 Astra at 76.2%, across 29 model configurations with a published result (Epoch AI Benchmarking Hub, result dated 26 Oct 2026). Questions that require reading and weighing research evidence to reach a supported conclusion. EBR-bench is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 6 Nov 2026)

  • GPT-6.1 Terra holds the top EBR-bench score at 78.1% (Epoch AI Benchmarking Hub, 26 Oct 2026).
  • GPT-6 Astra is second on EBR-bench at 76.2% (Epoch AI Benchmarking Hub, 3 Sep 2026).
  • EBR-bench: Evidence-based reasoning over research literature. Run by Epoch AI (Epoch AI Benchmarking Hub).
  • 29 model configurations have a published EBR-bench result on BenchLeader (BenchLeader, 6 Nov 2026).
  • EBR-bench does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Evidence-based reasoning over research literature. Run by Epoch AI.

Published by
Epoch AI Benchmarking Hub
Category
Knowledge
Index weight
Reference only
Models
29
Data as of
6 Nov 2026
Newest result
29 Oct 2026

CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.

What the test looks like

Questions that require reading and weighing research evidence to reach a supported conclusion.

How it is scored

Percent correct, run by Epoch AI.

What to keep in mind

Small set; new.

Top 15
  1. 1GPT-6.1 Terra (max)78.1%
  2. 2GPT-6 Astra (max)76.2%
  3. 3Claude Opus 5.5 (max)71.4%
  4. 4Gemini 4 Argon (high)64.8%
  5. 5Claude Fable 5.1 (max)57.1%
  6. 6GPT-6.1 Sol (max)54.3%
  7. 7GPT-6 Sol (max)53.3%
  8. 8Claude Opus 5 (max)45.7%
  9. 9GPT-5.6 Sol (max)44.8%
  10. 10Claude Haiku 5.5 (max)43.8%
  11. 11Claude Fable 5 (max)39.5%
  12. 12GPT-5.5 (xhigh)34.3%
  13. 13Muse Wonder (max)31.4%
  14. 14Grok 4.6 (xhigh)30.5%
  15. 15Claude Opus 4.8 (max)28.6%
29 of 29
#
1GPT-6.1 Terramax78.1%65.32026-10-26
2GPT-6 Astramax76.2%67.82026-09-03
3Claude Opus 5.5max71.4%69.32026-09-22
4Gemini 4 Argonhigh64.8%66.02026-09-30
5Claude Fable 5.1max57.1%67.22026-09-01
6GPT-6.1 Solmax54.3%67.32026-09-29
7GPT-6 Solmax53.3%63.32026-09-22
8Claude Opus 5max45.7%65.42026-07-24
9GPT-5.6 Solmax44.8%65.42026-07-09
10Claude Haiku 5.5max43.8%57.02026-10-07
11Claude Fable 5max39.5%65.32026-06-09
12GPT-5.5xhigh34.3%64.32026-04-23
13Muse Wondermax31.4%62.32026-10-12
14Grok 4.6xhigh30.5%57.92026-08-12
15Claude Opus 4.8max28.6%60.22026-05-28
16GPT-5.4xhigh25.4%63.22026-03-05
17GPT-5.2xhigh23.0%60.02025-12-11
18Claude Opus 4.7max19.1%59.52026-04-16
19NewMistral Large 4.0 OSShigh18.1%–2026-10-29
20Gemini 3.1 Pro14.3%60.52026-02-19
21Claude Opus 4.514.3%57.92025-11-24
22GPT-5high12.7%57.92025-08-07
23Claude Opus 4.6max12.7%55.82026-02-05
24Qwen3.7max9.5%60.32026-05-19
25GLM 5.2max9.5%59.52026-06-16
26Claude Opus 4.17.9%51.42025-08-05
27Gemini 3.5 Flashhigh4.8%62.52026-05-19
28Kimi K2.62.4%55.22026-04-20
29Claude Sonnet 4.52.4%53.72025-09-29

More knowledge benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on EBR-bench

Cite as: BenchLeader, “EBR-bench leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/ebr_bench.

EBR-bench: questions

What does EBR-bench measure?
Questions that require reading and weighing research evidence to reach a supported conclusion. Scores are reported in percent of tasks solved; higher is better.
What is the top score on EBR-bench?
As of 6 Nov 2026, the highest score on EBR-bench is 78.1% by GPT-6.1 Terra (result dated 26 Oct 2026), ahead of GPT-6 Astra at 76.2%. Scores come from Epoch AI Benchmarking Hub and are re-read daily.
Which AI model leads EBR-bench?
GPT-6.1 Terra leads EBR-bench with 78.1% as of 6 Nov 2026 (result dated 26 Oct 2026), ahead of GPT-6 Astra at 76.2%.
How many models have EBR-bench results?
29 model configurations have a EBR-bench result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
Who runs EBR-bench and how often is it updated?
EBR-bench is published by Epoch AI Benchmarking Hub. BenchLeader re-reads the published results once a day and records the date each result was published.
Does EBR-bench count toward the BenchLeader Index?
No. EBR-bench is shown for reference but left out of the composite index.