Skip to content
AI Atlas
BenchmarkActivecategory · compositefamily · artificial-analysis-intelligence-index · variant index

Artificial Analysis Intelligence Index

artificialanalysis.ai

composite of several evaluations run by Artificial Analysis

data quality57

Updated 4 h ago · first seen 11 Sept 2026

Metric
index ↑
Current results
1,272
Models
477
Current leader
Claude Fable 5.1 53.4

Score history · Claude Sonnet 5 12 rows

Score history for Claude Sonnet 5010203040Sept 26Sept 26Sept 26Sept 26Sept 26Sept 26Sept 26Sept 26Sept 26
  • Claude Sonnet 5
  • 24.74aa_slug=claude-sonnet-5-low · version=4.3 · estimated=false · reasoning_effort=low12 Sept 2026
  • 38.36aa_slug=claude-sonnet-5 · version=4.3 · estimated=false · reasoning_effort=max12 Sept 2026
  • 34.71aa_slug=claude-sonnet-5-xhigh · version=4.3 · estimated=false · reasoning_effort=xhigh12 Sept 2026
  • 32.03aa_slug=claude-sonnet-5-high · version=4.3 · estimated=false · reasoning_effort=high12 Sept 2026
  • 28.94aa_slug=claude-sonnet-5-non-reasoning · version=4.3 · estimated=true · reasoning=off12 Sept 2026
  • 28.44aa_slug=claude-sonnet-5-medium · version=4.3 · estimated=false · reasoning_effort=medium12 Sept 2026
  • 24.74aa_slug=claude-sonnet-5-low · version=4.3 · estimated=false · aa_variant_slug=claude-sonnet-5-low11 Sept 2026
  • 38.36aa_slug=claude-sonnet-5 · version=4.3 · estimated=false11 Sept 2026
  • 34.71aa_slug=claude-sonnet-5-xhigh · version=4.3 · estimated=false · aa_variant_slug=claude-sonnet-5-xhigh11 Sept 2026
  • 32.03aa_slug=claude-sonnet-5-high · version=4.3 · estimated=false · aa_variant_slug=claude-sonnet-5-high11 Sept 2026
  • 28.94aa_slug=claude-sonnet-5-non-reasoning · version=4.3 · estimated=true · reasoning=off11 Sept 2026
  • 28.44aa_slug=claude-sonnet-5-medium · version=4.3 · estimated=false · aa_variant_slug=claude-sonnet-5-medium11 Sept 2026

Back to the leaderboard

Frontier over time · index

8 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. 53.4Claude Fable 5.1 Anthropic Independent11 Sept 2026
  2. 53.2Claude Fable 5.1 Anthropic Independent11 Sept 2026
  3. 51.2Claude Fable 5.1 Anthropic Independent11 Sept 2026
  4. 51.0gpt-6-astra OpenAI Independent11 Sept 2026
  5. 49.7gpt-6-astra OpenAI Independent11 Sept 2026
  6. 45.1Claude Opus 5 Anthropic Independent11 Sept 2026
  7. 14.6grok-3-mini-reasoning SpaceXAI Independent11 Sept 2026
  8. 5.71llama-2-chat-7b Meta AI 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 477 models · trust independent-evaluator

Select models with +, then Compare.

Leaderboard
#ModelScoreTrustConfigurationvs leaderEvaluatedSourceActions
201Qwen3 VL 32B InstructOpen weightsQwen · Qwen3 · best of 4 rows11.9Independentreasoningonversion4.3Partially comparable0.00obs. 12 Sept 2026artificialanalysis.aiT2History
201ling-3-0-tinyOpen weightsinclusionAI · best of 2 rows11.9Independentreasoningonversion4.3Partially comparable0.00obs. 12 Sept 2026artificialanalysis.aiT2History
203Magistral Medium 1.2ClosedMistral AI · Magistral · best of 2 rows11.8Independentreasoningonversion4.3Partially comparable-0.04obs. 12 Sept 2026artificialanalysis.aiT2History
204Sonar Reasoning ProClosedPerplexity AI · Sonar · best of 2 rows11.8Independentreasoningonversion4.3Partially comparable-0.05obs. 12 Sept 2026artificialanalysis.aiT2History
205Granite 4.2 8BOpen weightsIBM · Granite 4.2 · best of 2 rows11.8Independentreasoningonversion4.3Partially comparable-0.06obs. 12 Sept 2026artificialanalysis.aiT2History
206hypernova-60bOpen weightsMultiverse Computing · best of 2 rows11.7Independentreasoning_efforthighversion4.3Partially comparable-0.13obs. 12 Sept 2026artificialanalysis.aiT2History
207MiniMax-M1-80kOpen weightsMiniMax · MiniMax · best of 2 rows11.7Independentreasoningonversion4.3Partially comparable-0.14obs. 12 Sept 2026artificialanalysis.aiT2History
208nemotron-cascade-2-30b-a3bOpen weightsNVIDIA · Nemotron · best of 2 rows11.7Independentreasoningonversion4.3Partially comparable-0.20obs. 12 Sept 2026artificialanalysis.aiT2History
209gemini-2-5-flash-reasoning-04-2025ClosedGoogle · Gemini 2.5 · best of 3 rows11.7Independentreasoningonversion4.3Partially comparable-0.22obs. 12 Sept 2026artificialanalysis.aiT2History
210Mercury 2ClosedInception · best of 2 rows11.5Independentreasoningonversion4.3Partially comparable-0.36obs. 12 Sept 2026artificialanalysis.aiT2History
211k2-think-v2Open weightsMBZUAI Institute of Foundation Models · best of 2 rows11.5Independentreasoningonversion4.3Partially comparable-0.37obs. 12 Sept 2026artificialanalysis.aiT2History
212longcat-flash-liteOpen weightsLongCat · best of 2 rows11.5Independentreasoningoffversion4.3Partially comparable-0.40obs. 12 Sept 2026artificialanalysis.aiT2History
213Mistral Small 4Open weightsMistral AI · Mistral · best of 4 rows11.4Independentreasoningonversion4.3Partially comparable-0.42obs. 12 Sept 2026artificialanalysis.aiT2History
214deepseek-r1-0120Open weightsDeepSeek · DeepSeek · best of 2 rows11.4Independentreasoningonversion4.3Partially comparable-0.46obs. 12 Sept 2026artificialanalysis.aiT2History
215o1 PreviewClosedOpenAI · OpenAI o-series · best of 2 rows11.4Independentreasoningonversion4.3Partially comparable-0.49obs. 12 Sept 2026artificialanalysis.aiT2History
216hyperclova-x-seed-think-32bOpen weightsNaver · Seed · best of 2 rows11.4Independentreasoningonversion4.3Partially comparable-0.51obs. 12 Sept 2026artificialanalysis.aiT2History
217GLM 4.6VOpen weightsZ.ai (Zhipu AI) · GLM4.6 · best of 4 rows11.2Independentreasoningonversion4.3Partially comparable-0.65obs. 12 Sept 2026artificialanalysis.aiT2History
218Qwen3 Next 80B A3B InstructOpen weightsQwen · Qwen3 · best of 4 rows11.2Independentreasoningonversion4.3Partially comparable-0.67obs. 12 Sept 2026artificialanalysis.aiT2History
219GLM 4.5 AirOpen weightsZ.ai (Zhipu AI) · GLM4.5 · best of 2 rows11.1Independentreasoningonversion4.3Partially comparable-0.78obs. 12 Sept 2026artificialanalysis.aiT2History
220mi-dm-k-2-5-pro-dec28ClosedKorea Telecom · best of 2 rows11.0Independentreasoningonversion4.3Partially comparable-0.83obs. 12 Sept 2026artificialanalysis.aiT2History
221ring-1tOpen weightsinclusionAI · best of 2 rows10.9Independentreasoningonversion4.3Partially comparable-0.97obs. 12 Sept 2026artificialanalysis.aiT2History
222Trinity Large ThinkingOpen weightsArcee AI · best of 2 rows10.9Independentreasoningonversion4.3Partially comparable-0.99obs. 12 Sept 2026artificialanalysis.aiT2History
223g9v3-3bOpen weightsAI9Stars · best of 2 rows10.8Independentreasoningonversion4.3Partially comparable-1.03obs. 12 Sept 2026artificialanalysis.aiT2History
224intellect-3Open weightsPrime Intellect · best of 2 rows10.6Independentreasoningonversion4.3Partially comparable-1.26obs. 12 Sept 2026artificialanalysis.aiT2History
225gpt-5-chatgptClosedOpenAI · GPT 5 · best of 2 rows10.4Independentreasoningoffversion4.3Partially comparable-1.43obs. 12 Sept 2026artificialanalysis.aiT2History
226solar-open-100b-reasoningOpen weightsUpstage · Solar · best of 2 rows10.4Independentreasoningonversion4.3Partially comparable-1.51obs. 12 Sept 2026artificialanalysis.aiT2History
227Gemini 2.5 Flash-LiteClosedGoogle · Gemini 2.5 · best of 5 rows10.3Independentreasoningonversion4.3Partially comparable-1.52obs. 12 Sept 2026artificialanalysis.aiT2History
227gemini-2-5-flash-lite-preview-09-2025ClosedGoogle · Gemini 2.5 · best of 3 rows10.3Independentreasoningonversion4.3Partially comparable-1.52obs. 11 Sept 2026artificialanalysis.aiT2History
229nemotron-3-nano-omni-30b-a3bOpen weightsNVIDIA · Nemotron 3 · best of 2 rows10.3Independentreasoningonversion4.3Partially comparable-1.62obs. 12 Sept 2026artificialanalysis.aiT2History
230gpt-4.1-miniClosedOpenAI · GPT 4.1 · best of 2 rows10.2Independentreasoningoffversion4.3Partially comparable-1.71obs. 12 Sept 2026artificialanalysis.aiT2History
231Qwen3 Coder NextOpen weightsQwen · Qwen3 · best of 2 rows10.1Independentreasoningoffversion4.3Partially comparable-1.82obs. 12 Sept 2026artificialanalysis.aiT2History
232MiniMax-M1-40kOpen weightsMiniMax · MiniMax · best of 2 rows9.99Independentreasoningonversion4.3Partially comparable-1.88obs. 12 Sept 2026artificialanalysis.aiT2History
233gpt-oss-20bOpen weightsOpenAI · gpt-oss · best of 4 rows9.95Independentreasoning_effortlowversion4.3Partially comparable-1.92obs. 12 Sept 2026artificialanalysis.aiT2History
234Mistral Medium 3.1ClosedMistral AI · Mistral · best of 2 rows9.89Independentreasoningoffversion4.3Partially comparable-1.98obs. 12 Sept 2026artificialanalysis.aiT2History
235k2-v2Open weightsMBZUAI Institute of Foundation Models · best of 6 rows9.87Independentreasoning_efforthighversion4.3Partially comparable-2.00obs. 12 Sept 2026artificialanalysis.aiT2History
236qwen3-30b-a3b-2507Open weightsAlibaba Group · Qwen3 · best of 4 rows9.81Independentreasoningonversion4.3Partially comparable-2.06obs. 12 Sept 2026artificialanalysis.aiT2History
237o1-miniClosedOpenAI · OpenAI o-series · best of 2 rows9.77Independentreasoningonversion4.3Partially comparable-2.10obs. 12 Sept 2026artificialanalysis.aiT2History
238DeepSeek V3 0324Open weightsDeepSeek · DeepSeek-V3 · best of 2 rows9.72Independentreasoningoffversion4.3Partially comparable-2.15obs. 12 Sept 2026artificialanalysis.aiT2History
239Mistral Large 3Open weightsMistral AI · Mistral · best of 2 rows9.71Independentreasoningoffversion4.3Partially comparable-2.16obs. 12 Sept 2026artificialanalysis.aiT2History
240ling-2-6-flashOpen weightsinclusionAI · best of 2 rows9.68Independentreasoningoffversion4.3Partially comparable-2.19obs. 12 Sept 2026artificialanalysis.aiT2History
241Qwen3 Coder 30B A3B InstructOpen weightsQwen · Qwen3 · best of 2 rows9.59Independentreasoningoffversion4.3Partially comparable-2.28obs. 12 Sept 2026artificialanalysis.aiT2History
241tri-21b-think-previewOpen weightsTrillion Labs · best of 2 rows9.59Independentreasoningonversion4.3Partially comparable-2.28obs. 12 Sept 2026artificialanalysis.aiT2History
243GPT-4.5 PreviewClosedOpenAI · GPT 4.5 · best of 2 rows9.58Independentreasoningoffversion4.3Partially comparable-2.29obs. 12 Sept 2026artificialanalysis.aiT2History
244diffusiongemma-26b-a4bOpen weightsGoogle · best of 2 rows9.51Independentreasoningonversion4.3Partially comparable-2.36obs. 12 Sept 2026artificialanalysis.aiT2History
245qwen3-235b-a22b-instructOpen weightsAlibaba Group · Qwen3 · best of 4 rows9.50Independentreasoningonversion4.3Partially comparable-2.37obs. 12 Sept 2026artificialanalysis.aiT2History
246QwQ-32BOpen weightsAlibaba Group · Qwen · best of 2 rows9.47Independentreasoningonversion4.3Partially comparable-2.40obs. 12 Sept 2026artificialanalysis.aiT2History
247Qwen3 VL 30B A3B InstructOpen weightsQwen · Qwen3 · best of 4 rows9.45Independentreasoningonversion4.3Partially comparable-2.42obs. 12 Sept 2026artificialanalysis.aiT2History
248gemini-2.0-flash-thinking-exp-01-21ClosedGoogle · Gemini 2.0 · best of 2 rows9.42Independentreasoningonversion4.3Partially comparable-2.45obs. 12 Sept 2026artificialanalysis.aiT2History
249Devstral 2Open weightsMistral AI · Devstral 2 · best of 2 rows9.41Independentreasoningoffversion4.3Partially comparable-2.46obs. 12 Sept 2026artificialanalysis.aiT2History
250Llama 4 MaverickOpen weightsMeta AI · Llama 4 · best of 2 rows9.30Independentreasoningoffversion4.3Partially comparable-2.57obs. 12 Sept 2026artificialanalysis.aiT2History
251motif-2-12-7bClosedMotif Technologies · best of 2 rows9.19Independentreasoningonversion4.3Partially comparable-2.68obs. 12 Sept 2026artificialanalysis.aiT2History
252ling-1tOpen weightsinclusionAI · best of 2 rows9.17Independentreasoningoffversion4.3Partially comparable-2.70obs. 12 Sept 2026artificialanalysis.aiT2History
253nova-premierClosedAmazon Web Services · Nova · best of 2 rows9.16Independentreasoningoffversion4.3Partially comparable-2.71obs. 12 Sept 2026artificialanalysis.aiT2History
254solar-pro-2ClosedUpstage · Solar · best of 5 rows9.07Independentreasoningonversion4.3Partially comparable-2.80obs. 12 Sept 2026artificialanalysis.aiT2History
254solar-pro-2-previewClosedUpstage · Solar · best of 3 rows9.07Independentreasoningonversion4.3Partially comparable-2.80obs. 11 Sept 2026artificialanalysis.aiT2History
256granite-4.2-3bOpen weightsIBM · Granite 4.2 · best of 2 rows9.06Independentreasoningonversion4.3Partially comparable-2.81obs. 12 Sept 2026artificialanalysis.aiT2History
257Mistral Medium 3ClosedMistral AI · Mistral · best of 2 rows9.05Independentreasoningoffversion4.3Partially comparable-2.82obs. 12 Sept 2026artificialanalysis.aiT2History
257magistral-mediumClosedMistral AI · Magistral · best of 2 rows9.05Independentreasoningonversion4.3Partially comparable-2.82obs. 12 Sept 2026artificialanalysis.aiT2History
259devstral-mediumClosedMistral AI · Devstral · best of 2 rows9.01Independentreasoningoffversion4.3Partially comparable-2.86obs. 12 Sept 2026artificialanalysis.aiT2History
259llama-nemotron-super-49b-v1-5Open weightsNVIDIA · Llama · best of 4 rows9.01Independentreasoningonversion4.3Partially comparable-2.86obs. 12 Sept 2026artificialanalysis.aiT2History
261tri-21b-think-v0-5Open weightsTrillion Labs · best of 2 rows8.99Independentreasoningonversion4.3Partially comparable-2.88obs. 12 Sept 2026artificialanalysis.aiT2History
262gpt-4o-chatgpt-03-25ClosedOpenAI · GPT 4 · best of 2 rows8.96Independentreasoningoffversion4.3Partially comparable-2.91obs. 12 Sept 2026artificialanalysis.aiT2History
263Claude Haiku 3.5ClosedAnthropic · Claude · best of 2 rows8.94Independentreasoningoffversion4.3Partially comparable-2.93obs. 12 Sept 2026artificialanalysis.aiT2History
263Gemini 2.0 FlashClosedGoogle · Gemini 2.0 · best of 2 rows8.94Independentreasoningoffversion4.3Partially comparable-2.93obs. 12 Sept 2026artificialanalysis.aiT2History
265llama-3-3-nemotron-super-49bOpen weightsNVIDIA · Llama 3.3 · best of 4 rows8.93Independentreasoningonversion4.3Partially comparable-2.94obs. 12 Sept 2026artificialanalysis.aiT2History
266gemma-4-E4BOpen weightsGoogle · Gemma 4 · best of 4 rows8.91Independentreasoningonversion4.3Partially comparable-2.96obs. 12 Sept 2026artificialanalysis.aiT2History
267Nemotron 3 Nano 30B A3BOpen weightsNVIDIA · Nemotron 3 · best of 4 rows8.90Independentreasoningonversion4.3Partially comparable-2.97obs. 12 Sept 2026artificialanalysis.aiT2History
268minicpm5-1bOpen weightsOpenBMB · best of 4 rows8.80Independentreasoningonversion4.3Partially comparable-3.07obs. 12 Sept 2026artificialanalysis.aiT2History
268qwen3-4b-2507-instructOpen weightsAlibaba Group · Qwen3 · best of 4 rows8.80Independentreasoningonversion4.3Partially comparable-3.07obs. 12 Sept 2026artificialanalysis.aiT2History
270sarvam-105bOpen weightsSarvam · best of 2 rows8.79Independentreasoning_efforthighversion4.3Partially comparable-3.08obs. 12 Sept 2026artificialanalysis.aiT2History
271gemini-2-0-pro-experimental-02-05ClosedGoogle · Gemini 2.0 · best of 2 rows8.75Independentreasoningoffversion4.3Partially comparable-3.12obs. 12 Sept 2026artificialanalysis.aiT2History
272Devstral Small 1.0Open weightsMistral AI · Devstral · best of 2 rows8.74Independentreasoningoffversion4.3Partially comparable-3.13obs. 12 Sept 2026artificialanalysis.aiT2History
273claude-3-opusClosedAnthropic · Claude 3 · best of 2 rows8.72Independentreasoningoffversion4.3Partially comparable-3.15obs. 12 Sept 2026artificialanalysis.aiT2History
274SonarClosedPerplexity AI · Sonar · best of 3 rows8.67Independentreasoningonversion4.3Partially comparable-3.20obs. 12 Sept 2026artificialanalysis.aiT2History
275gemini-2-5-flash-04-2025ClosedGoogle · Gemini 2.58.66Independentversion4.3Partially comparable-3.21obs. 11 Sept 2026artificialanalysis.aiT2History
276Magistral Small 1.2Open weightsMistral AI · Magistral · best of 2 rows8.59Independentreasoningonversion4.3Partially comparable-3.28obs. 12 Sept 2026artificialanalysis.aiT2History
277gpt-4oClosedOpenAI · GPT 4 · best of 2 rows8.44Independentreasoningoffversion4.3Partially comparable-3.43obs. 12 Sept 2026artificialanalysis.aiT2History
278nanbeige4-1-3bOpen weightsNanbeige · best of 2 rows8.40Independentreasoningonversion4.3Partially comparable-3.47obs. 12 Sept 2026artificialanalysis.aiT2History
279LFM2.5-2.6B (free)Open weightsLiquid AI · LFM2.5 · best of 2 rows8.39Independentreasoningonversion4.3Partially comparable-3.48obs. 12 Sept 2026artificialanalysis.aiT2History
280DeepSeek-R1-Distill-Qwen-32BOpen weightsDeepSeek · Qwen · best of 2 rows8.38Independentreasoningonversion4.3Partially comparable-3.49obs. 12 Sept 2026artificialanalysis.aiT2History
281Devstral Small 2Open weightsMistral AI · Devstral · best of 2 rows8.36Independentreasoningoffversion4.3Partially comparable-3.51obs. 12 Sept 2026artificialanalysis.aiT2History
282gemini-2.0-flash-expClosedGoogle · Gemini 2.0 · best of 2 rows8.22Independentreasoningoffversion4.3Partially comparable-3.65obs. 12 Sept 2026artificialanalysis.aiT2History
282magistral-smallOpen weightsMistral AI · Magistral · best of 2 rows8.22Independentreasoningonversion4.3Partially comparable-3.65obs. 12 Sept 2026artificialanalysis.aiT2History
284exaone-4-0-32bOpen weightsLG AI Research · EXAONE 4.0 · best of 4 rows8.18Independentreasoningonversion4.3Partially comparable-3.69obs. 12 Sept 2026artificialanalysis.aiT2History
285Qwen3 VL 8B InstructOpen weightsQwen · Qwen3 · best of 4 rows8.17Independentreasoningonversion4.3Partially comparable-3.70obs. 12 Sept 2026artificialanalysis.aiT2History
286DeepSeek-R1-0528-Qwen3-8BOpen weightsDeepSeek · Qwen3 · best of 2 rows8.08Independentreasoningonversion4.3Partially comparable-3.79obs. 12 Sept 2026artificialanalysis.aiT2History
287qwen-2-5-maxClosedAlibaba Group · Qwen · best of 2 rows8.02Independentreasoningoffversion4.3Partially comparable-3.85obs. 12 Sept 2026artificialanalysis.aiT2History
288Hermes-4-70BRestricted weightsNous Research · Hermes 47.91Independentversion4.3Partially comparable-3.96obs. 11 Sept 2026artificialanalysis.aiT2History
288gemini-1-5-proClosedGoogle · Gemini 1.5 · best of 2 rows7.91Independentreasoningoffversion4.3Partially comparable-3.96obs. 12 Sept 2026artificialanalysis.aiT2History
288hermes-4-llama-3-1-70bOpen weightsNous Research · Llama 3.1 · best of 3 rows7.91Independentreasoningonversion4.3Partially comparable-3.96obs. 12 Sept 2026artificialanalysis.aiT2History
291R1 Distill Llama 70BOpen weightsDeepSeek · Llama · best of 2 rows7.89Independentreasoningonversion4.3Partially comparable-3.98obs. 12 Sept 2026artificialanalysis.aiT2History
292claude-35-sonnetClosedAnthropic · Claude 35 · best of 2 rows7.88Independentreasoningoffversion4.3Partially comparable-3.99obs. 12 Sept 2026artificialanalysis.aiT2History
293deepseek-r1-distill-qwen-14bOpen weightsDeepSeek · Qwen · best of 2 rows7.85Independentreasoningonversion4.3Partially comparable-4.02obs. 12 Sept 2026artificialanalysis.aiT2History
294falcon-h1r-7bOpen weightsTII UAE · Falcon · best of 2 rows7.83Independentreasoningonversion4.3Partially comparable-4.04obs. 12 Sept 2026artificialanalysis.aiT2History
295Solar Pro 3ClosedUpstage · Solar · best of 2 rows7.82Independentreasoningonversion4.3Partially comparable-4.05obs. 12 Sept 2026artificialanalysis.aiT2History
295gpt-4.1-nanoClosedOpenAI · GPT 4.1 · best of 2 rows7.82Independentreasoningoffversion4.3Partially comparable-4.05obs. 12 Sept 2026artificialanalysis.aiT2History
297ling-flash-2-0Open weightsinclusionAI · best of 2 rows7.81Independentreasoningoffversion4.3Partially comparable-4.06obs. 12 Sept 2026artificialanalysis.aiT2History
298gemma-4-E2BOpen weightsGoogle · Gemma 4 · best of 4 rows7.77Independentreasoningonversion4.3Partially comparable-4.10obs. 12 Sept 2026artificialanalysis.aiT2History
299qwen3-omni-30b-a3b-instructOpen weightsAlibaba Group · Qwen3 · best of 4 rows7.76Independentreasoningonversion4.3Partially comparable-4.11obs. 12 Sept 2026artificialanalysis.aiT2History
300GPT-4o (2024-08-06)ClosedOpenAI · GPT 4 · best of 2 rows7.74Independentreasoningoffversion4.3Partially comparable-4.13obs. 12 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 →