Cleus
Start free
  1. Cleus
  2. AI benchmarks
  3. SimpleQA Verified

SimpleQA Verifiedexplained

Short facts, no web search, no making things up. SimpleQA Verified is 1,000 short questions that each have one indisputable answer, like a date, a name or a place. The model answers from memory with no web search. It is a cleaned-up version of OpenAI’s SimpleQA with bad and duplicate questions removed.

Facts · Google DeepMind, 2025 (based on OpenAI’s SimpleQA)

Highest published score
Gemini 4 Argon, 76.0%
Leader among popular models
Gemini 4 Argon
Models with a result
92
Skill it tests
Facts
01

What SimpleQA Verified really tests

What it is
SimpleQA Verified is 1,000 short questions that each have one indisputable answer, like a date, a name or a place. The model answers from memory with no web search. It is a cleaned-up version of OpenAI’s SimpleQA with bad and duplicate questions removed.
A question like
In what year did a particular regional museum first open to the public? (One right answer, easy to check, hard to know.) Our own example in the style of the test, not a real item.
Why it matters
This is the closest benchmark to “will it make things up?” Models that score higher both know more and are better at saying “I don’t know”.
Reading the score
Share of questions answered correctly.
The catch
It measures memory, not how good a model is once it can search the web. Most chat apps now search, so a low score matters less there.
Made by
Google DeepMind, 2025 (based on OpenAI’s SimpleQA). Official page
02

Leaderboard

Each model’s best published result across its reasoning settings.

Popular models

The models we track across this site.

  1. 1Gemini 4 ArgonGoogle76.0%
  2. 2GPT-6 AstraOpenAI75.6%
  3. 3GPT-6.1 SolOpenAI73.9%
  4. 4Claude Opus 5.5Anthropic72.2%
  5. 5Claude Fable 5.1Anthropic70.8%
  6. 6Gemini 3.8 FlashGoogle69.7%
  7. 7Kimi K3.1 ProMoonshot AI62.2%
  8. 8GPT-6.1 TerraOpenAI57.3%
  9. 9Muse Spark OpenMeta56.6%
  10. 10Grok 4.7xAI56.0%
  11. 11Kimi K3Moonshot AI50.6%
  12. 12Grok 4.6xAI49.3%
  13. 13Claude Sonnet 5.5Anthropic46.5%
  14. 14DeepSeek V4.1 FlashDeepSeek42.5%
  15. 15GLM-5.3Z.ai41.0%
  16. 16Claude Haiku 5.5Anthropic23.8%
  17. 17Mistral Large 4Mistral AI22.4%

Top 20 of all 92 models tested

Includes older, preview and specialist models that aren’t on our main leaderboard.

  1. 1Gemini 4 ArgonGoogle DeepMind · high76.0%
  2. 2GPT-6 AstraOpenAI · max75.6%
  3. 3GPT-6.1 SolOpenAI · max73.9%
  4. 4Gemini 3.1 ProGoogle DeepMind · high73.5%
  5. 5Claude Opus 5.5Anthropic · max72.2%
  6. 6Claude Fable 5.1Anthropic · max70.8%
  7. 7Claude Fable 5Anthropic · xhigh70.7%
  8. 8Gemini 3.8 FlashGoogle DeepMind · high69.7%
  9. 9GPT-5.6 SolOpenAI · max69.7%
  10. 10Gemini 3.7 FlashGoogle DeepMind · high69.2%
  11. 11Gemini 3 FlashGoogle DeepMind · high66.8%
  12. 12Gemini 3.5 FlashGoogle DeepMind · high66.2%
  13. 13Gemini 3.6 FlashGoogle DeepMind · high66.2%
  14. 14GPT-5.5OpenAI · xhigh63.0%
  15. 15Kimi K3.1 ProMoonshot · max62.2%
  16. 16GPT-6 SolOpenAI · max60.7%
  17. 17Muse Spark 1.2Meta AI · xhigh60.3%
  18. 18Claude Opus 5Anthropic · max59.9%
  19. 19Muse Spark 1.1Meta AI57.8%
  20. 20GPT-6.1 TerraOpenAI · max57.3%
03

Other facts tests

04

Questions

What is SimpleQA Verified?

SimpleQA Verified is 1,000 short questions that each have one indisputable answer, like a date, a name or a place. The model answers from memory with no web search. It is a cleaned-up version of OpenAI’s SimpleQA with bad and duplicate questions removed.

What does a SimpleQA Verified score mean?

Share of questions answered correctly.

Which AI model is best on SimpleQA Verified?

Gemini 4 Argon from Google DeepMind has the highest published SimpleQA Verified result we have (76.0%).

Is SimpleQA Verified a good benchmark?

This is the closest benchmark to “will it make things up?” Models that score higher both know more and are better at saying “I don’t know”. The catch: It measures memory, not how good a model is once it can search the web. Most chat apps now search, so a low score matters less there.

Who made SimpleQA Verified?

Google DeepMind, 2025 (based on OpenAI’s SimpleQA).

Benchmark scores: Epoch AI, Capabilities & benchmarking, used under CC BY 4.0. For each model we show its best published result across reasoning settings. Prices and context sizes: OpenRouter. Data refreshed November 9, 2026. Hands-on tests (answers, websites, code, refusals, speed) are our own single runs. Explanations written by Cleus.