Skip to content
IFAI

Model rankings

716 models · 4 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Text arena · Anonymous head-to-head votes · Published October 17, 2025

Most preferred
gemini-2.5-proGoogle1,451Score
Best open model
glm-4.6Z.ai1,422Score
Best from China
qwen3-max-previewAlibaba1,434Score
Most voted
claude-3-opus-20240229Anthropic196,368Votes
Top three by Score
1gemini-2.5-proGoogle1,451
Top three open models
1glm-4.6Z.ai · #13 overall1,422
2deepseek-v3.2-exp-thinkingDeepSeek · #16 overall1,419
3qwen3-vl-235b-a22b-instructAlibaba · #17 overall1,418
This week's changes
Magpie MultilingualSpeech · #17Up 11
Octave 2Speech · #20Up 5
Sonic English (Oct '24)Speech · #21Up 3
RankChangeModelScore95% intervalVotesPrice $/M
1—gemini-2.5-proClosedGoogle1,451±454,087—
2—claude-opus-4-1-20250805-thinking-16kClosedAnthropic1,447±521,306—
3—claude-sonnet-4-5-20250929-thinking-32kClosedAnthropic1,445±86,287—
4—gpt-4.5-preview-2025-02-27ClosedOpenAI1,441±614,644—
5—chatgpt-4o-latest-20250326ClosedOpenAI1,440±440,013—
6—o3-2025-04-16ClosedOpenAI1,440±451,293—
7—claude-sonnet-4-5-20250929ClosedAnthropic1,438±86,144—
8—gpt-5-highClosedOpenAI1,437±523,580—
9—claude-opus-4-1-20250805ClosedAnthropic1,437±533,298—
10—qwen3-max-previewClosedAlibaba1,434±618,078—
11—gpt-5-chatClosedOpenAI1,425±521,630—
12—qwen3-max-2025-09-23ClosedAlibaba1,423±76,919—
13—grok-4-fast-chatClosedxAI1,420±87,104—
14—claude-opus-4-20250514-thinking-16kClosedAnthropic1,419±535,522—
15—grok-4-0709ClosedxAI1,413±529,264—
16—claude-opus-4-20250514ClosedAnthropic1,411±443,310—
17—gpt-4.1-2025-04-14ClosedOpenAI1,411±441,918—
18—grok-3-preview-02-24ClosedxAI1,409±434,154—
19—gemini-2.5-flashClosedGoogle1,407±453,364—
20—mistral-medium-2508ClosedMistral · 2 more variants1,406±523,844—
21—gemini-2.5-flash-preview-09-2025ClosedGoogle1,404±76,730—
22—grok-4-fast-reasoningClosedxAI1,402±85,571—
23—o1-2024-12-17ClosedOpenAI1,400±428,039—
24—claude-sonnet-4-20250514-thinking-32kClosedAnthropic1,398±533,827—
25—claude-haiku-4-5-20251001ClosedAnthropic1,397±122,380$2.0
26—gpt-5-mini-highClosedOpenAI1,395±618,172—
27—o4-mini-2025-04-16ClosedOpenAI1,391±441,513—
28—hunyuan-vision-1.5-thinkingClosedTencent1,389±122,229—
29—claude-sonnet-4-20250514ClosedAnthropic1,389±539,329—
30—mai-1-previewClosedMicrosoft AI1,389±614,528—
31—o1-previewClosedOpenAI1,386±531,505—
32—claude-3-7-sonnet-20250219-thinking-32kClosedAnthropic1,386±439,987—
33—hunyuan-t1-20250711ClosedTencent1,384±94,845—
34—hunyuan-turbos-20250416ClosedTencent · 1 more variant1,379±611,135—
35—gpt-4.1-mini-2025-04-14ClosedOpenAI1,379±440,621—
36—gemini-2.5-flash-lite-preview-06-17-thinkingClosedGoogle1,374±531,701—
37—gemini-2.5-flash-lite-preview-09-2025-no-thinkingClosedGoogle1,374±76,765—
38—qwen2.5-maxClosedAlibaba1,372±433,541—
39—claude-3-5-sonnet-20241022ClosedAnthropic · 1 more variant1,370±389,889—
40—claude-3-7-sonnet-20250219ClosedAnthropic1,370±444,619—
41—grok-3-mini-highClosedxAI1,362±517,615—
42—o3-mini-highClosedOpenAI1,362±518,735—
43—gemini-2.0-flash-001ClosedGoogle1,360±445,135—
44—grok-3-mini-betaClosedxAI1,356±523,839—
45—gemini-2.0-flash-lite-preview-02-05ClosedGoogle1,352±425,215—
46—gemini-1.5-pro-002ClosedGoogle1,351±356,012—
47—o3-miniClosedOpenAI1,347±358,935—
48—amazon-nova-experimental-chat-10-09ClosedAmazon1,344±112,911—
49—gpt-4o-2024-05-13ClosedOpenAI1,344±3113,568—
50—qwen-plus-0125ClosedAlibaba1,343±85,861—

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 “—”.