Skip to content
IFAI

Model rankings

526 models · 3 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Intelligence Index · A composite of several evaluations · Updated September 10, 2025

Smartest
GPT-5 (high)OpenAI66.7Intelligence
Best open model
gpt-oss-120BOpenAI57.9Intelligence
Best value
Gemma 3n E4B InstructGoogle624 pts/$Value
Best from China
Qwen3 235B A22B 2507 (Reasoning)Alibaba57.5Intelligence
Top three by Intelligence
1GPT-5 (high)OpenAI66.7
2Grok 4xAI65.3
3o3-proOpenAI65.3
Top three open models
1gpt-oss-120BOpenAI · #9 overall57.9
2Qwen3 235B A22B 2507 (Reasoning)Alibaba · #10 overall57.5
3DeepSeek V3.1 (Reasoning)DeepSeek · #15 overall54.0
This week's changes
Grok Code Fast 1Intelligence · #22New
Qwen3 Max (Preview)Intelligence · #23New
Kimi K2 0905Intelligence · #19Up 12
RankChangeModelIntelligencePrice $/MSpeed tok/sFirst token
1▲ 1gpt-oss-120BOpenOpenAI57.9$0.26——
2▼ 1Qwen3 235B A22B 2507 (Reasoning)OpenAlibaba57.5$2.6——
3—DeepSeek V3.1 (Reasoning)OpenDeepSeek54.0$0.96——
4—DeepSeek R1 0528 (May '25)OpenDeepSeek52.0$0.96——
5NewKimi K2 0905OpenMoonshot AI50.4$1.4——
6▼ 1GLM-4.5OpenZ AI49.4$0.99——
7▲ 1GLM-4.5-AirOpenZ AI48.8$0.43——
8▲ 8Kimi K2OpenMoonshot AI48.2$1.1——
9▼ 3Qwen3 30B A3B 2507 (Reasoning)OpenAlibaba46.4$0.75——
10▼ 3MiniMax M1 80kOpenMiniMax46.2$0.83——
11▲ 1Qwen3 235B A22B 2507 (Non-reasoning)OpenAlibaba45.3$1.2——
12▼ 2Llama Nemotron Super 49B v1.5 (Reasoning)OpenNVIDIA45.2$0.000——
13▲ 2DeepSeek V3.1 (Non-reasoning)OpenDeepSeek44.8$0.48——
13▲ 1gpt-oss-20BOpenOpenAI44.8$0.088——
15▼ 6DeepSeek R1 (Jan '25)OpenDeepSeek43.8$1.8——
16▼ 5Qwen3 4B 2507 (Reasoning)OpenAlibaba43.4$0.000——
17▼ 4EXAONE 4.0 32B (Reasoning)OpenLG AI Research42.6$0.70——
18▲ 2Qwen3 Coder 480B A35BOpenAlibaba42.3$3.0——
19▼ 2Qwen3 235B A22B (Reasoning)OpenAlibaba41.7$2.6——
20▼ 1MiniMax M1 40kOpenMiniMax41.6$0.83——
21NewHermes 4 - Llama-3.1 405B (Reasoning)OpenNous Research41.6$1.5——
22NewDeepSeek V3 0324OpenDeepSeek41.3$0.48——
23NewHermes 4 - Llama-3.1 70B (Reasoning)OpenNous Research39.2$0.20——
24▲ 1GLM-4.5V (Reasoning)OpenZ AI38.8$0.85——
25▼ 3Qwen3 32B (Reasoning)OpenAlibaba38.7$2.6——
26▼ 8Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)OpenNVIDIA38.5$0.90——
27—NVIDIA Nemotron Nano 9B V2OpenNVIDIA38.1$0.000——
28—QwQ 32BOpenAlibaba37.9$0.63——
29▼ 8Qwen3 30B A3B 2507 (Non-reasoning)OpenAlibaba37.0$0.35——
30▼ 6NVIDIA Nemotron Nano 9B V2 (Reasoning)OpenNVIDIA36.9$0.000——
31▼ 5Qwen3 30B A3B (Reasoning)OpenAlibaba36.7$0.75——
32▼ 1Qwen3 14B (Reasoning)OpenAlibaba36.0$1.3——
33▼ 4Llama 4 MaverickOpenMeta35.8$0.39——
34▼ 2Llama 3.3 Nemotron Super 49B v1 (Reasoning)OpenNVIDIA35.5$0.000——
35▼ 2DeepSeek R1 0528 Qwen3 8BOpenDeepSeek34.8$0.065——
36▼ 2EXAONE 4.0 32BOpenLG AI Research33.4$0.70——
37▲ 2Qwen3 Coder 30B A3BOpenAlibaba33.4$0.90——
38▼ 3DeepSeek V3 (Dec '24)OpenDeepSeek33.2$0.48——
39▼ 3DeepSeek R1 Distill Qwen 32BOpenDeepSeek32.7$0.20——
40NewHermes 4 - Llama-3.1 405BOpenNous Research32.6$1.5——
41▼ 4Reka Flash 3OpenReka AI32.5$0.35——
42▼ 4Magistral SmallOpenMistral31.9$0.75——
43▼ 3DeepSeek R1 Distill Llama 70BOpenDeepSeek30.8$0.72——
44▼ 3Qwen3 235B A22BOpenAlibaba30.0$1.2——
45▼ 3DeepSeek R1 Distill Qwen 14BOpenDeepSeek29.7$0.88——
46▼ 1Mistral Small 3.2OpenMistral29.1$0.15——
47▼ 3Qwen3 14BOpenAlibaba28.4$0.61——
48▼ 18Qwen3 8B (Reasoning)OpenAlibaba28.3$0.66——
49▼ 6Llama 4 ScoutOpenMeta28.1$0.26——
50▼ 4Command AOpenCohere28.1$4.4——

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 “—”.