Skip to content
IFAI

Model rankings

245 models · 2 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Text arena · Anonymous head-to-head votes · Published February 27, 2025

Most preferred
o1-2024-12-17OpenAI1,323Score
Best open model
deepseek-r1DeepSeek1,320Score
Best from China
deepseek-r1DeepSeek1,320Score
Most voted
claude-3-opus-20240229Anthropic202,639Votes
Top three by Score
1o1-2024-12-17OpenAI1,323
2deepseek-r1DeepSeek1,320
Top three open models
1deepseek-r1DeepSeek · #2 overall1,320
2deepseek-v3DeepSeek · #9 overall1,277
3llama-3.1-405b-instruct-bf16Meta · #18 overall1,254
This week's changes
DeepSeek R1 Distill Qwen 14BIntelligence · #5New
DeepSeek R1 Distill Qwen 32BIntelligence · #10New
DeepSeek R1 Distill Llama 70BIntelligence · #13New
Qwen2.5 Instruct 72BIntelligence · #19New
Phi-4Intelligence · #20New
RankChangeModelScore95% intervalVotesPrice $/M
1—o1-2024-12-17ClosedOpenAI1,323±419,186—
2—deepseek-r1OpenDeepSeek1,320±78,099—
3Newclaude-3-7-sonnet-20250219ClosedAnthropic1,308±113,184—
4▼ 1o1-previewClosedOpenAI1,303±433,173—
5▼ 1o3-mini-highClosedOpenAI1,291±68,525—
6▼ 1gemini-2.0-flash-001ClosedGoogle1,288±612,616—
7▼ 1claude-3-5-sonnet-20241022ClosedAnthropic1,286±358,478—
8▼ 1qwen2.5-maxClosedAlibaba1,284±611,308—
9▼ 1deepseek-v3OpenDeepSeek1,277±421,425—
10—gemini-1.5-pro-002ClosedGoogle1,269±456,936—
11▼ 2o3-miniClosedOpenAI1,267±514,844—
12▼ 1gemini-2.0-flash-lite-preview-02-05ClosedGoogle1,266±612,173—
13▼ 1gpt-4o-2024-05-13ClosedOpenAI1,263±3117,723—
14▼ 1qwen-plus-0125ClosedAlibaba1,262±76,028—
15▼ 1claude-3-5-sonnet-20240620ClosedAnthropic1,260±386,126—
16▼ 1glm-4-plus-0111ClosedZhipu1,257±86,011—
17▼ 1o1-miniClosedOpenAI1,255±354,929—
18▼ 1gemini-advanced-0514ClosedGoogle1,254±352,134—
19▼ 1llama-3.1-405b-instruct-bf16OpenMeta1,254±433,602—
20▼ 1llama-3.1-405b-instruct-fp8OpenMeta1,253±363,130—
21▼ 1step-2-16k-exp-202412ClosedStepFun1,251±75,127—
22▼ 1gpt-4o-2024-08-06ClosedOpenAI1,250±347,960—
23▼ 1grok-2-2024-08-13ClosedxAI1,249±367,032—
24▼ 1yi-lightningClosed01 AI1,246±328,952—
25▼ 1gpt-4-turbo-2024-04-09ClosedOpenAI1,242±3102,112—
26▼ 1gemini-1.5-pro-001ClosedGoogle1,241±382,431—
27▼ 1deepseek-v2.5-1210OpenDeepSeek1,240±87,250—
28Newhunyuan-large-2025-02-10ClosedTencent1,240±93,635—
29▼ 2claude-3-opus-20240229ClosedAnthropic1,239±3202,639—
30▼ 2glm-4-plusClosedZhipu AI1,235±427,771—
31—llama-3.3-70b-instructOpenMeta1,235±427,913—
32▼ 3gpt-4o-mini-2024-07-18ClosedOpenAI1,235±366,391—
33▼ 3gpt-4-1106-previewClosedOpenAI1,235±3103,743—
34▼ 2claude-3-5-haiku-20241022ClosedAnthropic1,232±523,083—
35▼ 2mistral-large-2407OpenMistral1,232±348,181—
36▼ 2qwen2.5-plus-1127ClosedAlibaba1,231±610,731—
37▼ 2qwen-max-0919OpenAlibaba1,231±517,427—
38▼ 1athene-v2-chatOpenNexusFlow1,231±426,092—
39▼ 3gpt-4-0125-previewClosedOpenAI1,231±397,038—
40▼ 2gemini-1.5-flash-002ClosedGoogle1,230±436,985—
41Newhunyuan-standard-2025-02-10ClosedTencent1,226±93,792—
42▼ 3deepseek-v2.5OpenDeepSeek1,226±426,335—
43▼ 3grok-2-mini-2024-08-13ClosedxAI1,224±355,430—
44▼ 3athene-70b-0725OpenNexusFlow1,223±420,601—
45▼ 3qwen2.5-72b-instructOpenAlibaba1,222±441,527—
46▼ 3mistral-large-2411OpenMistral1,220±422,859—
47▼ 3llama-3.1-nemotron-70b-instructOpenNVIDIA1,216±87,599—
48▼ 3llama-3.1-70b-instructOpenMeta1,214±358,740—
49▼ 3amazon-nova-pro-v1.0ClosedAmazon1,209±419,052—
50▼ 3llama-3.1-tulu-3-70bOpenAi21,209±123,009—

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 “—”.