Skip to content
IFAI

Model rankings

443 models · 2 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Intelligence Index · A composite of several evaluations · Updated July 21, 2025

Smartest
Grok 4xAI73.2Intelligence
Best open model
DeepSeek R1 0528 (May '25)DeepSeek68.3Intelligence
Best value
Gemma 3n E4B InstructGoogle1,120 pts/$Value
Best from China
DeepSeek R1 0528 (May '25)DeepSeek68.3Intelligence
Top three by Intelligence
1Grok 4xAI73.2
2o3-proOpenAI71.0
3Gemini 2.5 ProGoogle70.5
Top three open models
1DeepSeek R1 0528 (May '25)DeepSeek · #7 overall68.3
2MiniMax M1 80kMiniMax · #12 overall63.0
3Qwen3 235B A22B (Reasoning)Alibaba · #14 overall62.3
This week's changes
Kimi K2Intelligence · #22New
Claude 4 Sonnet (Extended Thinking)Intelligence · #13Up 4
RankChangeModelIntelligencePrice $/MSpeed tok/sFirst token
1—DeepSeek R1 0528 (May '25)OpenDeepSeek · 1 more variant68.3$0.96——
2—MiniMax M1 80kOpenMiniMax63.0$0.83——
3—Qwen3 235B A22B (Reasoning)OpenAlibaba · 1 more variant62.3$2.6——
4—MiniMax M1 40kOpenMiniMax61.4$0.83——
5—Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)OpenNVIDIA60.8$0.90——
6—Qwen3 32B (Reasoning)OpenAlibaba · 1 more variant59.2$2.6——
7—QwQ 32BOpenAlibaba58.1$0.60——
8NewKimi K2OpenMoonshot AI57.6$1.3——
9—Qwen3 14B (Reasoning)OpenAlibaba · 1 more variant55.7$1.3——
10—Qwen3 30B A3B (Reasoning)OpenAlibaba · 1 more variant55.6$0.75——
11—Magistral SmallOpenMistral54.7$0.75——
12—DeepSeek V3 0324 (Mar '25)OpenDeepSeek · 1 more variant53.2$0.48——
13—DeepSeek R1 0528 Qwen3 8BOpenDeepSeek52.2$0.065——
14—DeepSeek R1 Distill Qwen 32BOpenDeepSeek51.9$0.30——
15—Qwen3 8B (Reasoning)OpenAlibaba · 1 more variant51.4$0.66——
16—Llama 3.3 Nemotron Super 49B v1 (Reasoning)OpenNVIDIA · 1 more variant51.2$0.000——
17—Llama 4 MaverickOpenMeta50.5$0.38——
18—DeepSeek R1 Distill Qwen 14BOpenDeepSeek49.0$0.20——
19—DeepSeek R1 Distill Llama 70BOpenDeepSeek48.2$0.72——
20—Qwen3 4B (Reasoning)OpenAlibaba · 1 more variant47.3$0.40——
21—Reka Flash 3OpenReka AI47.1$0.35——
22—Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning)OpenNVIDIA45.0$0.000——
23—Llama 4 ScoutOpenMeta43.0$0.24——
24—QwQ 32B-PreviewOpenAlibaba42.6$0.67——
25—Mistral Small 3.2OpenMistral42.3$0.15——
26—Llama 3.3 Instruct 70BOpenMeta41.1$0.62——
27—Llama 3.1 Instruct 405BOpenMeta40.5$3.3——
28—Qwen2.5 Instruct 72BOpenAlibaba40.4$0.000——
29—MiniMax-Text-01OpenMiniMax40.2$0.43——
30—Phi-4OpenMicrosoft Azure40.2$0.22——
31—Command AOpenCohere40.0$4.4——
32—Llama 3.1 Tulu3 405BOpenAllen Institute for AI39.9$0.000——
33—Mistral Large 2 (Nov '24)OpenMistral · 1 more variant38.3$3.0——
34—Qwen3 1.7B (Reasoning)OpenAlibaba · 1 more variant38.2$0.40——
35—Gemma 3 27B InstructOpenGoogle37.6$0.000——
36—Pixtral LargeOpenMistral37.4$3.0——
37—Qwen2.5 Instruct 32BOpenAlibaba37.4$0.15——
38—Llama 3.1 Nemotron Instruct 70BOpenNVIDIA37.3$0.16——
39—Qwen2.5 Coder Instruct 32BOpenAlibaba36.3$0.15——
40—Llama 3.1 Instruct 70BOpenMeta35.4$0.76——
41—Mistral Small 3.1OpenMistral35.3$0.15——
42—Mistral Small 3OpenMistral35.3$0.15——
43—DeepSeek-V2.5 (Dec '24)OpenDeepSeek · 1 more variant35.3$0.18——
44—Devstral Small (May '25)OpenMistral · 1 more variant34.1$0.15——
45—DeepSeek R1 Distill Llama 8BOpenDeepSeek34.0$0.040——
46—Gemma 3 12B InstructOpenGoogle33.8$0.063——
47—R1 1776OpenPerplexity33.7$3.5——
48—Llama 3.2 Instruct 90B (Vision)OpenMeta33.4$0.54——
48—Solar MiniOpenUpstage33.4$0.15——
50—Grok-1OpenxAI32.7$0.000——

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 “—”.