BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases9

SimpleQA Verified

Short factual questions; measures recall without hallucination. Run by Epoch AI.

As of 25 Oct 2026, Gemini 4 Argon leads SimpleQA Verified on BenchLeader with 76.0%, ahead of GPT-6 Astra at 75.6%, across 87 model configurations with a published result.

Published by
Epoch AI Benchmarking Hub
Category
Knowledge
Index weight
1.0
Models
87
Data as of
25 Oct 2026
Newest result
17 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

Short factual questions with a single verifiable answer ("Who won the 1988 Nobel Prize in Chemistry?"), drawn from a cleaned and re-checked subset of OpenAI's SimpleQA.

How it is scored

Percent answered correctly; a model can also abstain. Epoch AI runs and grades the answers automatically against the reference.

What to keep in mind

It measures recall of stored facts, not reasoning, and it favours large models with broad pretraining. A model that refuses more often lowers its score but may hallucinate less; read it alongside a hallucination-aware test such as AA-Omniscience.

Top 15
  1. 1Gemini 4 Argon (high)76.0%
  2. 2GPT-6 Astra (max)75.6%
  3. 3GPT-6.1 Sol (max)73.9%
  4. 4Gemini 3.1 Pro (high)73.5%
  5. 5Claude Opus 5.5 (max)72.2%
  6. 6Claude Fable 5.1 (max)70.8%
  7. 7Claude Fable 5 (xhigh)70.7%
  8. 8GPT-5.6 Sol (max)69.7%
  9. 9Gemini 3.8 Flash (high)69.7%
  10. 10Gemini 3.7 Flash (high)69.2%
  11. 11Gemini 3 Flash (high)66.8%
  12. 12Gemini 3.5 Flash (high)66.2%
  13. 13Gemini 3.6 Flash (high)66.2%
  14. 14GPT-5.5 (xhigh)63.0%
  15. 15GPT-6 Sol (max)60.7%
87 of 87
#
1Gemini 4 Argonhigh76.0%70.72026-09-30
2GPT-6 Astramax75.6%70.42026-09-03
3GPT-6.1 Solmax73.9%69.92026-09-29
4Gemini 3.1 Prohigh73.5%60.32026-02-19
5Claude Opus 5.5max72.2%72.62026-09-22
6Claude Fable 5.1max70.8%70.52026-09-01
7Claude Fable 5xhigh70.7%–2026-06-09
8GPT-5.6 Solmax69.7%68.02026-07-09
9Gemini 3.8 Flashhigh69.7%64.42026-09-02
10Gemini 3.7 Flashhigh69.2%63.52026-08-13
11Gemini 3 Flashhigh66.8%58.32025-12-17
12Gemini 3.5 Flashhigh66.2%63.12026-05-19
13Gemini 3.6 Flashhigh66.2%62.02026-07-21
14GPT-5.5xhigh63.0%66.12026-04-23
15GPT-6 Solmax60.7%66.42026-09-22
16Muse Spark 1.2xhigh60.3%63.22026-08-05
17Claude Opus 5max59.9%68.82026-07-24
18Muse Spark 1.157.8%62.52026-07-09
19Muse Spark Open 816B A42Bxhigh56.6%58.62026-10-15
20Grok 4.7xhigh56.0%61.62026-09-21
21Qwen3.7max55.8%60.62026-05-19
22Claude Opus 4.8max53.0%63.32026-05-28
23DeepSeek V4 Promax52.9%60.82026-08-13
24Qwen3.6 Maxmax52.0%62.32026-04-20
25Claude Opus 4.7xhigh51.7%57.42026-04-16
26Kimi K3max50.6%64.82026-07-16
27GPT-5high50.1%57.92025-08-07
28o3high49.4%53.92025-04-16
29Grok 4.6high49.3%62.92026-08-12
30Grok 4.6xhigh48.9%62.52026-08-12
31Qwen3 Maxmax48.8%52.82025-09-24
32Grok 4.5high48.3%59.32026-07-08
33GPT-5.1high48.0%57.72025-11-13
34Qwen3.8 Maxxhigh47.3%57.12026-09-01
35Claude Opus 4.6max47.0%57.42026-02-05
36Claude Sonnet 5.5max46.5%69.42026-09-28
37GPT-5.4 Proxhigh46.3%61.92026-03-05
38Claude Opus 4.545.7%57.92025-11-24
39GPT-5.4xhigh45.1%64.72026-03-05
40Qwen3.6 Plus44.1%58.82026-03-31
41GPT-5.6 Terramax43.2%64.42026-07-09
42DeepSeek V4.1 Flashmax42.9%61.22026-09-09
43DeepSeek V4.1 Flash 1017max42.5%64.02026-10-17
44GPT-6 Lunamax41.4%60.02026-09-22
45o1high41.1%48.32024-12-17
46GLM 5.3max41.0%63.52026-08-14
47GPT-5.6 Lunamax41.0%59.92026-07-09
48Qwen3 235B A22B 2507thinking40.4%50.32025-07-25
49GPT-5.2xhigh37.1%60.72025-12-11
50Kimi K2.7 Code36.5%55.82026-06-12
51Claude Sonnet 4.6high35.5%50.72026-02-17
52Kimi K2.634.9%55.32026-04-20
53GPT-5.2high34.3%56.02025-12-11
54Kimi K2.534.3%54.52026-01-27
55GLM 5.2max34.2%62.32026-06-16
56GLM 5.134.0%55.92026-04-07
57Claude Sonnet 5max33.7%58.42026-06-30
58DeepSeek V4 Flashmax33.6%59.42026-07-31
59Grok 4.3high33.2%57.02026-04-17
60Claude Sonnet 5xhigh32.9%59.02026-06-30

More knowledge benchmarks · All LLM benchmarks

Cite as: BenchLeader, “SimpleQA Verified leaderboard”, data as of 25 Oct 2026, https://www.benchleader.com/benchmarks/simpleqa_verified.

SimpleQA Verified: questions

What does SimpleQA Verified measure?
Short factual questions with a single verifiable answer ("Who won the 1988 Nobel Prize in Chemistry?"), drawn from a cleaned and re-checked subset of OpenAI's SimpleQA. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads SimpleQA Verified?
Gemini 4 Argon leads SimpleQA Verified with 76.0% as of 25 Oct 2026, ahead of GPT-6 Astra at 75.6%.
How many models have SimpleQA Verified results?
87 model configurations have a SimpleQA Verified result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
Who runs SimpleQA Verified and how often is it updated?
SimpleQA Verified 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 SimpleQA Verified count toward the BenchLeader Index?
Yes. SimpleQA Verified contributes to the knowledge category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.