Skip to content
AI Atlas
BenchmarkActivecategory · reasoningfamily · gpqa · variant main

GPQA

github.com/idavidrein/gpqa

graduate-level science questions (main set, 448 questions)

quality57

Updated 3 min ago · first seen 11 Sept 2026

Metric
accuracy · % ↑
Current results
614
Models
607
Current leader
—

Score history · qwen3-vl-8b-reasoning 1 row

Not enough history to chart — a single observation (57.88% on 11 Sept 2026). Rows under different configurations count separately; the list below shows each one.

  • 57.88%aa_slug=qwen3-vl-8b-reasoning · variant=GPQA Diamond · evaluator=Artificial Analysis · index_version=4.311 Sept 2026

Back to the leaderboard

Frontier over time · accuracy · variant=GPQA Diamond · evaluator=Artificial Analysis

9 leader changes recorded, all dated 11 Sept 2026 — the frontier line needs at least two distinct dates. The corpus is young: every result was first observed on the same day, so leader changes will separate in time as sources are re-crawled.

  1. 96.3%gpt-6-astra-xhigh OpenAI Independent11 Sept 2026
  2. 96.1%gpt-6-astra OpenAI Independent11 Sept 2026
  3. 95.3%Gemini 3.8 Flash Google Independent11 Sept 2026
  4. 95.0%gpt-6-astra-high OpenAI Independent11 Sept 2026
  5. 93.9%gpt-6-astra-medium OpenAI Independent11 Sept 2026
  6. 93.5%Kimi K3 Moonshot AI Independent11 Sept 2026
  7. 91.9%claude-opus-5-medium Anthropic Independent11 Sept 2026
  8. 79.1%grok-3-mini-reasoning SpaceXAI Independent11 Sept 2026

Includes closed rows (history). A point is emitted whenever a result beats every earlier result of the same group, ordered by evaluated_at when the source publishes it, else observed_at.

Leaderboard 607 models

Select models with +, then Compare.

Leaderboard
#ModelScoreTrustConfigurationvs leaderEvaluatedSourceActions
301deepseek-v4-pro-0424-non-reasoningOpen weightsDeepSeek71.7%Independentgroup defaultsComparable0.00 ptobs. 11 Sept 2026artificialanalysis.aiT2History
302deepseek-v4-flash-0420-non-reasoningOpen weightsDeepSeek71.6%Independentgroup defaultsComparable-0.10 ptobs. 11 Sept 2026artificialanalysis.aiT2History
303gemma-4-26b-a4b-non-reasoningOpen weightsGoogle71.4%Independentgroup defaultsComparable-0.31 ptobs. 11 Sept 2026artificialanalysis.aiT2History
304apriel-v1-5-15b-thinkerOpen weightsServiceNow71.3%Independentgroup defaultsComparable-0.41 ptobs. 11 Sept 2026artificialanalysis.aiT2History
304k2-think-v2Open weightsMBZUAI Institute of Foundation Models71.3%Independentgroup defaultsComparable-0.41 ptobs. 11 Sept 2026artificialanalysis.aiT2History
306Qwen3 VL 235B A22B InstructOpen weightsQwen71.2%Independentgroup defaultsComparable-0.51 ptobs. 11 Sept 2026artificialanalysis.aiT2History
306gpt-5-2-non-reasoningClosedOpenAI71.2%Independentgroup defaultsComparable-0.51 ptobs. 11 Sept 2026artificialanalysis.aiT2History
306qwen3-5-4b-non-reasoningOpen weightsAlibaba Group71.2%Independentgroup defaultsComparable-0.51 ptobs. 11 Sept 2026artificialanalysis.aiT2History
309gemini-2-5-flash-lite-preview-09-2025-reasoningClosedGoogle70.9%Independentgroup defaultsComparable-0.81 ptobs. 11 Sept 2026artificialanalysis.aiT2History
310deepseek-r1-0120Open weightsDeepSeek70.8%Independentgroup defaultsComparable-0.91 ptobs. 11 Sept 2026artificialanalysis.aiT2History
311qwen3-30b-a3b-2507-reasoningOpen weightsAlibaba Group70.7%Independentgroup defaultsComparable-1.01 ptobs. 11 Sept 2026artificialanalysis.aiT2History
312minicpm5-2bOpen weightsOpenBMB70.2%Independentgroup defaultsComparable-1.52 ptobs. 11 Sept 2026artificialanalysis.aiT2History
313claude-4-opusClosedAnthropic70.1%Independentgroup defaultsComparable-1.62 ptobs. 11 Sept 2026artificialanalysis.aiT2History
313gemini-2.0-flash-thinking-exp-01-21ClosedGoogle70.1%Independentgroup defaultsComparable-1.62 ptobs. 11 Sept 2026artificialanalysis.aiT2History
313mi-dm-k-2-5-pro-dec28ClosedKorea Telecom70.1%Independentgroup defaultsComparable-1.62 ptobs. 11 Sept 2026artificialanalysis.aiT2History
316qwen3-235b-a22b-instruct-reasoningOpen weightsAlibaba Group70%Independentgroup defaultsComparable-1.72 ptobs. 11 Sept 2026artificialanalysis.aiT2History
317Hermes-4-70BOpen weightsNous Research69.9%Independentgroup defaultsComparable-1.82 ptobs. 11 Sept 2026artificialanalysis.aiT2History
317nova-2-0-omni-reasoning-lowClosedAmazon Web Services69.9%Independentgroup defaultsComparable-1.82 ptobs. 11 Sept 2026artificialanalysis.aiT2History
319gemini-2-5-flash-reasoning-04-2025ClosedGoogle69.8%Independentgroup defaultsComparable-1.92 ptobs. 11 Sept 2026artificialanalysis.aiT2History
319nova-2-0-lite-reasoning-lowClosedAmazon Web Services69.8%Independentgroup defaultsComparable-1.92 ptobs. 11 Sept 2026artificialanalysis.aiT2History
321MiniMax-M1-80kOpen weightsMiniMax69.7%Independentgroup defaultsComparable-2.02 ptobs. 11 Sept 2026artificialanalysis.aiT2History
322Qwen3 VL 30B A3B InstructOpen weightsQwen69.5%Independentgroup defaultsComparable-2.23 ptobs. 11 Sept 2026artificialanalysis.aiT2History
322k-exaone-non-reasoningOpen weightsLG AI Research69.5%Independentgroup defaultsComparable-2.23 ptobs. 11 Sept 2026artificialanalysis.aiT2History
322motif-2-12-7bClosedMotif Technologies69.5%Independentgroup defaultsComparable-2.23 ptobs. 11 Sept 2026artificialanalysis.aiT2History
325grok-3ClosedSpaceXAI69.3%Independentgroup defaultsComparable-2.43 ptobs. 11 Sept 2026artificialanalysis.aiT2History
326step-3-vl-10bOpen weightsStepFun69.0%Independentgroup defaultsComparable-2.73 ptobs. 11 Sept 2026artificialanalysis.aiT2History
327gpt-oss-20bOpen weightsOpenAI68.8%Independentgroup defaultsComparable-2.93 ptobs. 11 Sept 2026artificialanalysis.aiT2History
328gpt-5-mini-minimalClosedOpenAI68.7%Independentgroup defaultsComparable-3.03 ptobs. 11 Sept 2026artificialanalysis.aiT2History
328solar-pro-2-reasoningClosedUpstage68.7%Independentgroup defaultsComparable-3.03 ptobs. 11 Sept 2026artificialanalysis.aiT2History
330glm-5-2-non-reasoningOpen weightsZ.ai (Zhipu AI)68.6%Independentgroup defaultsComparable-3.13 ptobs. 11 Sept 2026artificialanalysis.aiT2History
330gpt-5-chatgptClosedOpenAI68.6%Independentgroup defaultsComparable-3.13 ptobs. 11 Sept 2026artificialanalysis.aiT2History
332GLM 4.5VOpen weightsZ.ai (Zhipu AI) · best of 2 rows68.4%Independentgroup defaultsComparable-3.34 ptobs. 11 Sept 2026artificialanalysis.aiT2History
333Claude Sonnet 4ClosedAnthropic · Claude68.3%Independentgroup defaultsComparable-3.44 ptobs. 11 Sept 2026artificialanalysis.aiT2History
333Gemini 2.5 FlashClosedGoogle · Gemini68.3%Independentgroup defaultsComparable-3.44 ptobs. 11 Sept 2026artificialanalysis.aiT2History
335MiniMax-M1-40kOpen weightsMiniMax68.2%Independentgroup defaultsComparable-3.54 ptobs. 11 Sept 2026artificialanalysis.aiT2History
336k2-v2Open weightsMBZUAI Institute of Foundation Models68.1%Independentgroup defaultsComparable-3.64 ptobs. 11 Sept 2026artificialanalysis.aiT2History
337Mistral Large 3Open weightsMistral AI68.0%Independentgroup defaultsComparable-3.74 ptobs. 11 Sept 2026artificialanalysis.aiT2History
338magistral-mediumClosedMistral AI67.9%Independentgroup defaultsComparable-3.84 ptobs. 11 Sept 2026artificialanalysis.aiT2History
339gpt-5-nanoClosedOpenAI67.6%Independentgroup defaultsComparable-4.14 ptobs. 11 Sept 2026artificialanalysis.aiT2History
339jt-miniClosedChina Mobile67.6%Independentgroup defaultsComparable-4.14 ptobs. 11 Sept 2026artificialanalysis.aiT2History
341gpt-5-minimalClosedOpenAI67.3%Independentgroup defaultsComparable-4.45 ptobs. 11 Sept 2026artificialanalysis.aiT2History
342claude-4-5-haiku-reasoningClosedAnthropic67.2%Independentgroup defaultsComparable-4.55 ptobs. 11 Sept 2026artificialanalysis.aiT2History
342gpt-oss-120b-lowOpen weightsOpenAI67.2%Independentgroup defaultsComparable-4.55 ptobs. 11 Sept 2026artificialanalysis.aiT2History
344Llama 4 MaverickOpen weightsMeta AI67.1%Independentgroup defaultsComparable-4.65 ptobs. 11 Sept 2026artificialanalysis.aiT2History
344Qwen3 VL 32B InstructOpen weightsQwen67.1%Independentgroup defaultsComparable-4.65 ptobs. 11 Sept 2026artificialanalysis.aiT2History
346gpt-5-nano-mediumClosedOpenAI67.0%Independentgroup defaultsComparable-4.75 ptobs. 11 Sept 2026artificialanalysis.aiT2History
347diffusiongemma-26b-a4bOpen weightsGoogle66.9%Independentgroup defaultsComparable-4.85 ptobs. 11 Sept 2026artificialanalysis.aiT2History
348Qwen3 32BOpen weightsQwen66.8%Independentgroup defaultsComparable-4.95 ptobs. 11 Sept 2026artificialanalysis.aiT2History
349qwen3-4b-2507-instruct-reasoningOpen weightsAlibaba Group66.7%Independentgroup defaultsComparable-5.05 ptobs. 11 Sept 2026artificialanalysis.aiT2History
350glm-5-non-reasoningOpen weightsZ.ai (Zhipu AI)66.6%Independentgroup defaultsComparable-5.15 ptobs. 11 Sept 2026artificialanalysis.aiT2History
350gpt-4.1ClosedOpenAI66.6%Independentgroup defaultsComparable-5.15 ptobs. 11 Sept 2026artificialanalysis.aiT2History
352glm-4-7-non-reasoningOpen weightsZ.ai (Zhipu AI)66.4%Independentgroup defaultsComparable-5.36 ptobs. 11 Sept 2026artificialanalysis.aiT2History
352gpt-4.1-miniClosedOpenAI66.4%Independentgroup defaultsComparable-5.36 ptobs. 11 Sept 2026artificialanalysis.aiT2History
354Magistral Small 1.2Open weightsMistral AI66.3%Independentgroup defaultsComparable-5.46 ptobs. 11 Sept 2026artificialanalysis.aiT2History
355falcon-h1r-7bOpen weightsTII UAE66.1%Independentgroup defaultsComparable-5.66 ptobs. 11 Sept 2026artificialanalysis.aiT2History
355gemma-4-12b-non-reasoningOpen weightsGoogle66.1%Independentgroup defaultsComparable-5.66 ptobs. 11 Sept 2026artificialanalysis.aiT2History
357qwen3-30b-a3b-2507Open weightsAlibaba Group65.9%Independentgroup defaultsComparable-5.86 ptobs. 11 Sept 2026artificialanalysis.aiT2History
358grok-4-3-non-reasoningClosedSpaceXAI65.8%Independentgroup defaultsComparable-5.96 ptobs. 11 Sept 2026artificialanalysis.aiT2History
359ling-flash-2-0Open weightsinclusionAI65.7%Independentgroup defaultsComparable-6.06 ptobs. 11 Sept 2026artificialanalysis.aiT2History
359solar-open-100b-reasoningOpen weightsUpstage65.7%Independentgroup defaultsComparable-6.06 ptobs. 11 Sept 2026artificialanalysis.aiT2History
361Claude 3.7 SonnetClosedAnthropic · Claude65.6%Independentgroup defaultsComparable-6.16 ptobs. 11 Sept 2026artificialanalysis.aiT2History
361mimo-v2-flashOpen weightsXiaomi65.6%Independentgroup defaultsComparable-6.16 ptobs. 11 Sept 2026artificialanalysis.aiT2History
363DeepSeek V3 0324Open sourceDeepSeek · DeepSeek-V365.5%Independentgroup defaultsComparable-6.27 ptobs. 11 Sept 2026artificialanalysis.aiT2History
363gpt-4o-chatgpt-03-25ClosedOpenAI65.5%Independentgroup defaultsComparable-6.27 ptobs. 11 Sept 2026artificialanalysis.aiT2History
365gemini-2-5-flash-lite-preview-09-2025ClosedGoogle65.0%Independentgroup defaultsComparable-6.67 ptobs. 11 Sept 2026artificialanalysis.aiT2History
366Claude Haiku 4.5ClosedAnthropic · Claude64.7%Independentgroup defaultsComparable-7.07 ptobs. 11 Sept 2026artificialanalysis.aiT2History
367gpt-5-6-luna-non-reasoningClosedOpenAI64.5%Independentgroup defaultsComparable-7.17 ptobs. 11 Sept 2026artificialanalysis.aiT2History
368granite-4.2-30bOpen weightsIBM64.4%Independentgroup defaultsComparable-7.28 ptobs. 11 Sept 2026artificialanalysis.aiT2History
369gpt-5-1-non-reasoningClosedOpenAI64.3%Independentgroup defaultsComparable-7.38 ptobs. 11 Sept 2026artificialanalysis.aiT2History
369llama-3-3-nemotron-super-49b-reasoningOpen weightsNVIDIA64.3%Independentgroup defaultsComparable-7.38 ptobs. 11 Sept 2026artificialanalysis.aiT2History
371magistral-smallOpen weightsMistral AI64.1%Independentgroup defaultsComparable-7.58 ptobs. 11 Sept 2026artificialanalysis.aiT2History
372grok-4-1-fastClosedSpaceXAI63.7%Independentgroup defaultsComparable-7.98 ptobs. 11 Sept 2026artificialanalysis.aiT2History
373gemini-2.0-flash-expClosedGoogle63.6%Independentgroup defaultsComparable-8.08 ptobs. 11 Sept 2026artificialanalysis.aiT2History
373longcat-flash-liteOpen weightsLongCat63.6%Independentgroup defaultsComparable-8.08 ptobs. 11 Sept 2026artificialanalysis.aiT2History
373nova-2-0-proClosedAmazon Web Services63.6%Independentgroup defaultsComparable-8.08 ptobs. 11 Sept 2026artificialanalysis.aiT2History
376sarvam-30bOpen weightsSarvam63.3%Independentgroup defaultsComparable-8.39 ptobs. 11 Sept 2026artificialanalysis.aiT2History
377GLM 4.6Open weightsZ.ai (Zhipu AI)63.2%Independentgroup defaultsComparable-8.49 ptobs. 11 Sept 2026artificialanalysis.aiT2History
378Granite 4.2 8BOpen weightsIBM63.1%Independentgroup defaultsComparable-8.59 ptobs. 11 Sept 2026artificialanalysis.aiT2History
378celeris-1ClosedCeleris63.1%Independentgroup defaultsComparable-8.59 ptobs. 11 Sept 2026artificialanalysis.aiT2History
380exaone-4-0-32bOpen weightsLG AI Research62.8%Independentgroup defaultsComparable-8.89 ptobs. 11 Sept 2026artificialanalysis.aiT2History
381Gemini 2.5 Flash-LiteClosedGoogle · Gemini · best of 2 rows62.5%Independentgroup defaultsComparable-9.19 ptobs. 11 Sept 2026artificialanalysis.aiT2History
382Gemini 2.0 FlashClosedGoogle · Gemini62.3%Independentgroup defaultsComparable-9.40 ptobs. 11 Sept 2026artificialanalysis.aiT2History
382sonar-reasoningClosedPerplexity AI62.3%Independentgroup defaultsComparable-9.40 ptobs. 11 Sept 2026artificialanalysis.aiT2History
384gemini-2-0-pro-experimental-02-05ClosedGoogle62.2%Independentgroup defaultsComparable-9.50 ptobs. 11 Sept 2026artificialanalysis.aiT2History
385qwen3-omni-30b-a3b-instructOpen weightsAlibaba Group62.0%Independentgroup defaultsComparable-9.70 ptobs. 11 Sept 2026artificialanalysis.aiT2History
386qwen3-coder-480b-a35b-instructOpen weightsAlibaba Group · Qwen3-Coder61.8%Independentgroup defaultsComparable-9.90 ptobs. 11 Sept 2026artificialanalysis.aiT2History
387qwen3-30b-a3b-instruct-reasoningOpen weightsAlibaba Group61.6%Independentgroup defaultsComparable-10.1 ptobs. 11 Sept 2026artificialanalysis.aiT2History
388DeepSeek-R1-Distill-Qwen-32BOpen weightsDeepSeek61.5%Independentgroup defaultsComparable-10.2 ptobs. 11 Sept 2026artificialanalysis.aiT2History
388hyperclova-x-seed-think-32bOpen weightsNaver61.5%Independentgroup defaultsComparable-10.2 ptobs. 11 Sept 2026artificialanalysis.aiT2History
390qwen3-235b-a22b-instructOpen weightsAlibaba Group61.3%Independentgroup defaultsComparable-10.4 ptobs. 11 Sept 2026artificialanalysis.aiT2History
391DeepSeek-R1-0528-Qwen3-8BOpen weightsDeepSeek61.2%Independentgroup defaultsComparable-10.5 ptobs. 11 Sept 2026artificialanalysis.aiT2History
392gpt-oss-20b-lowOpen weightsOpenAI61.1%Independentgroup defaultsComparable-10.6 ptobs. 11 Sept 2026artificialanalysis.aiT2History
393olmo-3-32b-thinkOpen weightsAllen Institute for AI61.0%Independentgroup defaultsComparable-10.7 ptobs. 11 Sept 2026artificialanalysis.aiT2History
394gpt-5-4-mini-non-reasoningClosedOpenAI60.6%Independentgroup defaultsComparable-11.1 ptobs. 11 Sept 2026artificialanalysis.aiT2History
394grok-4-fastClosedSpaceXAI60.6%Independentgroup defaultsComparable-11.1 ptobs. 11 Sept 2026artificialanalysis.aiT2History
396Qwen3 14BOpen weightsQwen60.4%Independentgroup defaultsComparable-11.3 ptobs. 11 Sept 2026artificialanalysis.aiT2History
397nova-2-0-liteClosedAmazon Web Services60.3%Independentgroup defaultsComparable-11.4 ptobs. 11 Sept 2026artificialanalysis.aiT2History
397o1-miniClosedOpenAI60.3%Independentgroup defaultsComparable-11.4 ptobs. 11 Sept 2026artificialanalysis.aiT2History
399tri-21b-think-v0-5Open weightsTrillion Labs60.1%Independentgroup defaultsComparable-11.6 ptobs. 11 Sept 2026artificialanalysis.aiT2History
400claude-35-sonnetClosedAnthropic59.9%Independentgroup defaultsComparable-11.8 ptobs. 11 Sept 2026artificialanalysis.aiT2History

One row per canonical model — its best current row inside this comparability group (effort variants are folded into the model). Bars are relative to the page's best score. “vs leader” reads comparability: partially comparable = same task, conditions differ (reasoning effort, temperature, judge). Rules →