// BENCHMARK

SimpleQA benchmark

AI model leaderboard for the SimpleQA benchmark. Compare how large language models score on SimpleQA, see the full ranking, and understand what this AI benchmark measures. Tongyi DeepResearch currently leads with 98.6%. Short-answer factuality benchmark with 4,326 fact-seeking questions.

Leaderboard

# Model Organization Score Variant Source
#1 Tongyi DeepResearch Alibaba-NLP 98.6% with-search official ↗
#2 DeepSeek-V3.2-Exp DeepSeek 97.1% from-hf-readme official ↗
#3 gemini-2.5-pro Google DeepMind 54% official ↗
#4 Grok 4 xAI 51.9% official ↗
#5 GPT-4.1 OpenAI 42.3% official ↗
#6 GPT-4o (Nov 2024) OpenAI 39% cited-minimax-text official ↗
#7 Kimi K2 Moonshot 31% official ↗
#8 Claude 3.5 Sonnet Anthropic 28.1% 1022-cited-minimax-text official ↗
#9 DeepSeek-R1-0528 DeepSeek 27.8% official ↗
#10 DeepSeek-V3-0324 DeepSeek 27.7% official ↗
#11 Gemini 2.0 Flash Google DeepMind 26.6% exp-cited-minimax-text official ↗
#12 GLM 4.5 Zhipu AI 26.4% official ↗
#13 DeepSeek-V3 DeepSeek 24.9% cited-minimax-text official ↗
#14 DeepSeek-V3 DeepSeek 24.6% official ↗
#15 MiniMax-Text-01 MiniMax 23.7% from-hf-readme official ↗
#16 Gemini 1.5 Pro Google DeepMind 23.4% 002-cited-minimax-text official ↗
#17 Llama 3.1-405B Meta AI 23.2% instruct-cited-minimax-text official ↗
#18 Claude Sonnet 4 Anthropic 18.5% official ↗
#19 MiniMax-M1-80k MiniMax 18.5% official ↗
#20 MiniMax-M1-40k MiniMax 17.9% official ↗
#21 GLM-4.5-Air Zhipu AI 14.5% official ↗
#22 Mistral Small 3.2 Mistral AI 12.1% instruct-totalacc official ↗
#23 Mistral Small 3.1 Mistral AI 10.43% instruct-totalacc official ↗
#24 Qwen2.5 Instruct (72B) Qwen 10.3% instruct-cited-minimax-text official ↗
#25 dots.llm1 Rednote 9.3% official ↗
#26 IBM Granite 4.1 IBM 4.82%

Frequently asked questions about SimpleQA

What is the SimpleQA benchmark?

Short-answer factuality benchmark with 4,326 fact-seeking questions.

How is the SimpleQA benchmark scored?

SimpleQA is scored using the accuracy (%) metric, where a higher score is better. The maximum achievable score is 100.000. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.

Which AI model scores highest on SimpleQA?

As of the latest reported scores on GenAIList, Tongyi DeepResearch achieves the highest result on SimpleQA with a score of 98.6%.

Is a higher SimpleQA score better?

Yes. On SimpleQA a higher score indicates better performance, so models near the top of the leaderboard are the strongest.