Skip to content
IFAI

Model rankings

532 models · 3 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Text arena · Anonymous head-to-head votes · Published September 18, 2025

Most preferred
gemini-2.5-proGoogle1,456Score
Best open model
Best from China
qwen3-max-previewAlibaba1,430Score
Most voted
claude-3-opus-20240229Anthropic196,368Votes
Top three by Score
1gemini-2.5-proGoogle1,456
3o3-2025-04-16OpenAI1,441
Top three open models
1qwen3-235b-a22b-instruct-2507Alibaba · #12 overall1,419
2kimi-k2-0711-previewMoonshot AI · #13 overall1,418
3deepseek-r1-0528DeepSeek · #14 overall1,418
This week's changes
Qwen3 Next 80B A3B (Reasoning)Intelligence · #13New
RankChangeModelScore95% intervalVotesPrice $/M
1—gemini-2.5-proClosedGoogle1,456±546,291—
2—claude-opus-4-1-20250805-thinking-16kClosedAnthropic1,449±614,843—
3—o3-2025-04-16ClosedOpenAI1,441±446,083—
4▲ 1chatgpt-4o-latest-20250326ClosedOpenAI1,441±535,807—
5▲ 1gpt-4.5-preview-2025-02-27ClosedOpenAI1,441±614,644—
6▼ 2gpt-5-highClosedOpenAI1,440±617,636—
7—claude-opus-4-1-20250805ClosedAnthropic1,438±623,612—
8▲ 1qwen3-max-previewClosedAlibaba1,430±711,851—
9▼ 1gpt-5-chatClosedOpenAI1,430±614,876—
10▲ 1claude-opus-4-20250514-thinking-16kClosedAnthropic1,421±530,101—
11▼ 1grok-4-0709ClosedxAI1,420±523,994—
12▲ 2gpt-4.1-2025-04-14ClosedOpenAI1,411±536,527—
13—grok-3-preview-02-24ClosedxAI1,410±435,154—
14▼ 2claude-opus-4-20250514ClosedAnthropic1,410±538,104—
15—mistral-medium-2508ClosedMistral · 2 more variants1,407±616,052—
16—gemini-2.5-flashClosedGoogle1,406±547,004—
17—claude-sonnet-4-20250514-thinking-32kClosedAnthropic1,399±528,637—
18—o1-2024-12-17ClosedOpenAI1,399±428,039—
19—o4-mini-2025-04-16ClosedOpenAI1,392±536,291—
20—mai-1-previewClosedMicrosoft AI1,391±79,951—
21—gpt-5-mini-highClosedOpenAI1,389±712,733—
22▲ 1claude-sonnet-4-20250514ClosedAnthropic1,386±534,172—
23▼ 1hunyuan-t1-20250711ClosedTencent1,385±94,915—
24—o1-previewClosedOpenAI1,385±531,505—
25—claude-3-7-sonnet-20250219-thinking-32kClosedAnthropic1,385±436,956—
26▲ 1hunyuan-turbos-20250416ClosedTencent · 1 more variant1,380±611,597—
27▼ 1gemini-2.5-flash-lite-preview-06-17-thinkingClosedGoogle1,378±526,274—
28—gpt-4.1-mini-2025-04-14ClosedOpenAI1,375±536,061—
29—qwen2.5-maxClosedAlibaba1,371±434,240—
30—claude-3-7-sonnet-20250219ClosedAnthropic1,369±441,863—
31—claude-3-5-sonnet-20241022ClosedAnthropic · 1 more variant1,368±386,848—
32—grok-3-mini-highClosedxAI1,362±614,820—
33▲ 1gemini-2.0-flash-001ClosedGoogle1,362±442,303—
34▼ 1o3-mini-highClosedOpenAI1,361±518,735—
35—grok-3-mini-betaClosedxAI1,357±521,142—
36—gemini-2.0-flash-lite-preview-02-05ClosedGoogle1,351±425,215—
37—gemini-1.5-pro-002ClosedGoogle1,350±356,012—
38—o3-miniClosedOpenAI1,346±454,461—
39▲ 2gpt-5-nano-highClosedOpenAI1,344±94,775—
40▼ 1qwen-plus-0125ClosedAlibaba1,343±85,861—
41▼ 1gpt-4o-2024-05-13ClosedOpenAI1,342±3113,568—
42—glm-4-plus-0111ClosedZhipu · 1 more variant1,342±85,806—
43—hunyuan-turbo-0110ClosedTencent1,337±112,322—
44—o1-miniClosedOpenAI1,333±452,301—
45—gpt-4o-2024-08-06ClosedOpenAI1,332±445,787—
46—grok-2-2024-08-13ClosedxAI1,332±463,725—
47▲ 1step-2-16k-exp-202412ClosedStepFun1,331±94,895—
48▼ 1gemini-advanced-0514ClosedGoogle1,331±550,654—
49—yi-lightningClosed01 AI1,326±527,624—
50—hunyuan-large-2025-02-10ClosedTencent1,322±103,760—

Intelligence over time

Each maker's best model at the time on the Artificial Analysis Intelligence Index, by the month it was released.

Gained most · Anthropic +57.2Highest now · Anthropic 64.6Gap between highest and lowest 18.1 ptsTime range Nov 2025 – Nov 2026
64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi

Arena data comes from the leaderboard dataset it publishes on Hugging Face (CC BY 4.0). Only each board's overall leaderboard is used; scores and ranks are unchanged. dataset · CC BY 4.0

Artificial Analysis data comes from its API (artificialanalysis.ai). Scores are unchanged; ranks follow the scores. artificialanalysis.ai

Open or closed follows the license Arena gives; prices and Artificial Analysis intelligence show only where model names match, otherwise “—”.