Skip to content
IFAI

Model rankings

391 models · 2 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Text arena · Anonymous head-to-head votes · Published June 1, 2025

Most preferred
o3-2025-04-16OpenAI1,442Score
Best open model
deepseek-v3-0324DeepSeek1,386Score
Best from China
deepseek-v3-0324DeepSeek1,386Score
Most voted
claude-3-opus-20240229Anthropic202,641Votes
Top three by Score
1o3-2025-04-16OpenAI1,442
Top three open models
1deepseek-v3-0324DeepSeek · #10 overall1,386
2deepseek-r1DeepSeek · #11 overall1,382
3gemma-3-27b-itGoogle · #22 overall1,351
This week's changes
Claude 4 OpusIntelligence · #14New
Grok 3 Reasoning BetaIntelligence · #16New
o1-proIntelligence · #16New
GPT-4.5 (Preview)Intelligence · #22New
RankChangeModelScore95% intervalVotesPrice $/M
1—o3-2025-04-16ClosedOpenAI1,442±513,133—
2—chatgpt-4o-latest-20250326ClosedOpenAI1,429±417,656—
3—gpt-4.5-preview-2025-02-27ClosedOpenAI1,424±515,271—
4Newclaude-opus-4-20250514ClosedAnthropic1,414±67,729—
5▼ 1gpt-4.1-2025-04-14ClosedOpenAI1,399±611,773—
6▼ 1grok-3-preview-02-24ClosedxAI1,397±419,977—
7Newclaude-sonnet-4-20250514ClosedAnthropic1,390±76,384—
8▼ 1o4-mini-2025-04-16ClosedOpenAI1,389±511,452—
9▼ 3o1-2024-12-17ClosedOpenAI1,386±329,038—
10▼ 1o1-previewClosedOpenAI1,370±333,177—
11▼ 3claude-3-7-sonnet-20250219-thinking-32kClosedAnthropic1,364±417,342—
12▼ 2claude-3-7-sonnet-20250219ClosedAnthropic1,364±323,874—
13▼ 2gpt-4.1-mini-2025-04-14ClosedOpenAI1,363±510,582—
14▼ 1claude-3-5-sonnet-20241022ClosedAnthropic · 1 more variant1,354±271,805—
15▼ 1gemini-2.0-flash-001ClosedGoogle1,353±332,757—
16▼ 4o3-mini-highClosedOpenAI1,353±419,404—
17▼ 2gemma-3-27b-itOpenGoogle1,351±418,283—
18▼ 2gemini-2.0-flash-lite-preview-02-05ClosedGoogle1,337±326,104—
19▼ 2o3-miniClosedOpenAI1,337±330,374—
20▼ 2gemini-1.5-pro-002ClosedGoogle1,336±358,645—
21▼ 1gpt-4o-2024-05-13ClosedOpenAI1,331±2117,747—
22▼ 1gemma-3-12b-itOpenGoogle1,328±73,976—
23▼ 1o1-miniClosedOpenAI1,323±354,951—
24▼ 1gemini-advanced-0514ClosedGoogle1,321±252,144—
25▼ 1llama-3.1-405b-instruct-bf16OpenMeta1,321±243,788—
26▼ 1llama-3.1-405b-instruct-fp8OpenMeta1,320±263,038—
27▼ 1gpt-4o-2024-08-06ClosedOpenAI1,318±347,973—
28—llama-4-maverick-17b-128e-instructOpenMeta1,317±513,745—
29▼ 2grok-2-2024-08-13ClosedxAI1,317±267,084—
30▼ 1llama-3.3-nemotron-49b-super-v1OpenNVIDIA1,314±92,371—
31Newgrok-3-mini-betaClosedxAI1,310±84,764—
32▼ 1gpt-4-turbo-2024-04-09ClosedOpenAI1,310±2102,133—
33▼ 1gemini-1.5-pro-001ClosedGoogle1,308±282,435—
34▼ 4gpt-4.1-nano-2025-04-14ClosedOpenAI1,307±76,302—
35▼ 2claude-3-opus-20240229ClosedAnthropic1,306±2202,641—
36—gpt-4-1106-previewClosedOpenAI1,302±2103,748—
37▼ 3gpt-4o-mini-2024-07-18ClosedOpenAI1,302±272,536—
38▼ 3llama-3.3-70b-instructOpenMeta1,302±343,364—
39▼ 2claude-3-5-haiku-20241022ClosedAnthropic1,301±342,779—
40▼ 1gpt-4-0125-previewClosedOpenAI1,298±297,079—
41—gemini-1.5-flash-002ClosedGoogle1,297±337,021—
42▼ 4gemma-3n-e4b-itOpenGoogle1,296±83,913—
43▼ 1grok-2-mini-2024-08-13ClosedxAI1,292±355,442—
44▼ 1gemma-3-4b-itOpenGoogle1,288±94,321—
45▼ 1llama-3.1-nemotron-70b-instructOpenNVIDIA1,284±77,577—
46▼ 1llama-3.1-70b-instructOpenMeta1,281±258,637—
47▼ 1llama-3.1-tulu-3-70bOpenAi21,277±93,010—
48▼ 1amazon-nova-pro-v1.0ClosedAmazon1,276±326,371—
49▼ 1gemma-2-27b-itOpenGoogle1,275±279,538—
50▼ 1gemini-1.5-flash-001ClosedGoogle1,272±265,661—

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 “—”.