Benchmark · Reasoning and knowledge

SimpleQA Verified

1,000 short factual questions with no answer choices, made by Google from OpenAI's SimpleQA. Epoch AI asks the model to guess rather than abstain, has a grader model mark each answer and reports the share answered correctly, not Google's F1 score.

Score
% of questions correct, higher is better
Run by
Epoch AI
Status
Current, newest result 19 November 2026
Updated
as models are released
Data
the published file, CC BY 4.0, read 20 November 2026
Method
epoch.ai/benchmarks/simple-qa-verified

Results

30 models we list, each at its best setting, best first.

#ModelScoreRangeWho ran itSettingDate
1Claude Fable 5.2 A94.0%Independenteffort max2026-11-06source
2GPT-6.2 Astra AA78.5%Independenteffort max2026-10-27source
3Gemini 4 Carbon B77.7%Independenteffort high2026-11-02source
4Gemini 4 Argon B76.0%Independenteffort high2026-10-13source
5GPT-6 Astra A75.6%Independenteffort max2026-08-30source
6Gemini 4 Neon B75.2%Independenteffort xhigh2026-11-19source
7GPT-6.1 Sol AA73.9%Independenteffort max2026-09-29source
8Gemini 3.1 Pro (preview) AA73.5%Independenteffort high2026-08-10source
9Claude Opus 5.5 A72.2%Independenteffort max2026-09-22source
10Gemini 4 Hydrogen B70.9%Independenteffort xhigh2026-11-13source
11Claude Fable 5.1 A70.8%Independenteffort max2026-09-01source
12Gemini 3.8 Flash AA69.7%Independenteffort high2026-09-02source
13Kimi K3.1 Pro D62.2%Independenteffort max2026-11-06source
14GPT-6 Sol A60.7%Independenteffort max2026-09-22source
15GPT-6.1 Terra AA57.3%Independenteffort max2026-10-26source
16Muse Spark Open C56.6%Independenteffort xhigh2026-10-15source
17Grok 4.7 BB56.0%Independenteffort xhigh2026-09-22source
18Kimi K3 B50.6%Independenteffort max2026-08-27source
19DeepSeek V4 Pro D47.0%Independenteffort max2026-08-27source
20Claude Sonnet 5.5 A46.5%Independenteffort max2026-09-29source
21DeepSeek V4.1 Flash C42.9%Independenteffort max2026-10-13source
22GPT-6 Luna A41.4%Independenteffort max2026-09-22source
23GLM-5.3 B41.0%Independenteffort max2026-08-28source
24Inkling B40.3%Independenteffort xhigh2026-08-27source
25DeepSeek-V4-Flash-0731 C33.6%Independenteffort max2026-08-27source
26Grok 4.3 BB33.2%Independenteffort high2026-08-27source
27Claude Haiku 5.5 A23.8%Independenteffort max2026-10-09source
28Mistral Large 4 C22.4%Independenteffort high2026-11-02source
29Inkling-Small B19.1%Independenteffort xhigh2026-08-27source
30Gemma 4 31B IT BB10.4%Independent2026-08-27source
Every setting, 31 rows
ModelAs the leaderboard names itScoreWho ran itSettingDate
Claude Fable 5.2claude-fable-5-294.0%Independenteffort max2026-11-06source
GPT-6.2 Astragpt-6.2-astra78.5%Independenteffort max2026-10-27source
Gemini 4 Carbongemini-4-carbon77.7%Independenteffort high2026-11-02source
Gemini 4 Argongemini-4-argon76.0%Independenteffort high2026-10-13source
GPT-6 Astragpt-6-astra75.6%Independenteffort max2026-08-30source
Gemini 4 Neongemini-4-neon75.2%Independenteffort xhigh2026-11-19source
GPT-6.1 Solgpt-6.1-sol73.9%Independenteffort max2026-09-29source
Gemini 3.1 Pro (preview)gemini-3.1-pro-preview73.5%Independenteffort high2026-08-10source
Claude Opus 5.5claude-opus-5-572.2%Independenteffort max2026-09-22source
Gemini 4 Hydrogengemini-4-hydrogen70.9%Independenteffort xhigh2026-11-13source
Claude Fable 5.1claude-fable-5-170.8%Independenteffort max2026-09-01source
Gemini 3.8 Flashgemini-3.8-flash69.7%Independenteffort high2026-09-02source
Kimi K3.1 Prokimi-k3.1-pro-together62.2%Independenteffort max2026-11-06source
GPT-6 Solgpt-6-sol60.7%Independenteffort max2026-09-22source
GPT-6.1 Terragpt-6.1-terra57.3%Independenteffort max2026-10-26source
Muse Spark OpenMuse-Spark-Open-816B-A42B56.6%Independenteffort xhigh2026-10-15source
Grok 4.7grok-4.756.0%Independenteffort xhigh2026-09-22source
Kimi K3kimi-k350.6%Independenteffort max2026-08-27source
DeepSeek V4 Prodeepseek-v4-pro47.0%Independenteffort max2026-08-27source
Claude Sonnet 5.5claude-sonnet-5-546.5%Independenteffort max2026-09-29source
DeepSeek V4.1 Flashdeepseek-v4.1-flash42.9%Independenteffort max2026-10-13source
DeepSeek V4.1 Flashdeepseek-v4.1-flash-101742.5%Independenteffort max2026-10-19source
GPT-6 Lunagpt-6-luna41.4%Independenteffort max2026-09-22source
GLM-5.3glm-5.341.0%Independenteffort max2026-08-28source
InklingInkling40.3%Independenteffort xhigh2026-08-27source
DeepSeek-V4-Flash-0731deepseek-v4-flash-073133.6%Independenteffort max2026-08-27source
Grok 4.3grok-4.333.2%Independenteffort high2026-08-27source
Claude Haiku 5.5claude-haiku-5-523.8%Independenteffort max2026-10-09source
Mistral Large 4mistral-large-4-0-oss22.4%Independenteffort high2026-11-02source
Inkling-Smallinkling-small19.1%Independenteffort xhigh2026-08-27source
Gemma 4 31B ITgemma-4-31b-it10.4%Independent2026-08-27source

A rank is among the models we list with a result, not a place on the whole leaderboard, and a model offered by several hosts is shown once, on its own listing. These results don't move our grades. Data: Epoch AI, Capabilities and benchmarking (CC BY 4.0), CC BY 4.0.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.