Skip to content
IFAI

Model rankings

880 models · 5 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Intelligence Index · A composite of several evaluations · Updated December 16, 2025

Smartest
Gemini 3 Pro Preview (high)Google72.9Intelligence
Best open model
Kimi K2 ThinkingKimi67.0Intelligence
Best value
Gemma 3n E4B InstructGoogle620 pts/$Value
Best from China
Kimi K2 ThinkingKimi67.0Intelligence
Top three by Intelligence
2GPT-5.2 (xhigh)OpenAI72.6
Top three open models
1Kimi K2 ThinkingKimi · #7 overall67.0
2DeepSeek V3.2 (Reasoning)DeepSeek · #9 overall65.9
3MiniMax-M2MiniMax · #19 overall61.4
This week's changes
GPT-5.2 (xhigh)Intelligence · #2New
gemini-3-flashCoding · #4New
deepseek-v3.2-thinkingCoding · #12New
Apriel-v1.6-15B-ThinkerIntelligence · #31New
RankChangeModelIntelligencePrice $/MSpeed tok/sFirst token
1—Gemini 3 Pro Preview (high)ClosedGoogle72.9$4.5——
2NewGPT-5.2 (xhigh)ClosedOpenAI72.6$4.8——
3▼ 1Claude Opus 4.5 (Reasoning)ClosedAnthropic69.8$10——
4▼ 1GPT-5.1 (high)ClosedOpenAI69.7$3.4——
5▼ 1GPT-5 Codex (high)ClosedOpenAI68.5$3.4——
6▼ 1GPT-5 (high)ClosedOpenAI68.5$3.4——
7▼ 1Kimi K2 ThinkingOpenKimi67.0$1.1——
8▼ 1GPT-5.1 Codex (high)ClosedOpenAI66.9$3.4——
9▼ 1GPT-5 (medium)ClosedOpenAI66.4$3.4——
10▼ 1DeepSeek V3.2 (Reasoning)OpenDeepSeek65.9$0.32——
11▼ 1o3ClosedOpenAI65.5$3.5——
12▼ 1Grok 4ClosedxAI65.3$6.0——
13▼ 1o3-proClosedOpenAI65.3$35——
14▼ 1Gemini 3 Pro Preview (low)ClosedGoogle64.5$4.5——
15▼ 1GPT-5 mini (high)ClosedOpenAI64.3$0.69——
16▼ 1Grok 4.1 Fast (Reasoning)ClosedxAI64.1$0.28——
17▼ 1KAT-Coder-Pro V1ClosedKwaiKAT63.7$0.000——
18▼ 1Claude 4.5 Sonnet (Reasoning)ClosedAnthropic62.7$6.0——
19▼ 1Nova 2.0 Pro Preview (medium)ClosedAmazon62.4$3.4——
20▼ 1GPT-5.1 Codex mini (high)ClosedOpenAI62.3$0.69——
21▼ 1GPT-5 (low)ClosedOpenAI61.8$3.4——
22▼ 1MiniMax-M2OpenMiniMax61.4$0.52——
23▼ 1GPT-5 mini (medium)ClosedOpenAI60.8$0.69——
24▼ 1gpt-oss-120B (high)OpenOpenAI60.5$0.26——
25▼ 1Grok 4 Fast (Reasoning)ClosedxAI60.3$0.28——
26▼ 1Claude Opus 4.5 (Non-reasoning)ClosedAnthropic59.9$10——
27▼ 1Gemini 2.5 ProClosedGoogle59.6$3.4——
28▼ 1o4-mini (high)ClosedOpenAI59.6$1.9——
29▼ 1Claude 4.1 Opus (Reasoning)ClosedAnthropic59.3$30——
30▼ 1DeepSeek V3.2 SpecialeOpenDeepSeek58.6$0.32——
31▼ 1Nova 2.0 Lite (medium)ClosedAmazon57.7$0.85——
32▼ 1DeepSeek V3.1 Terminus (Reasoning)OpenDeepSeek57.7$0.80——
33▼ 1Nova 2.0 Pro Preview (low)ClosedAmazon57.6$3.4——
34▼ 1Qwen3 235B A22B 2507 (Reasoning)OpenAlibaba57.5$2.6——
35▼ 1Doubao Seed CodeClosedByteDance Seed57.2$0.41——
36▼ 1Grok 3 mini Reasoning (high)ClosedxAI57.1$0.35——
37NewApriel-v1.6-15B-ThinkerOpenServiceNow57.0$0.000——
38▼ 2DeepSeek V3.2 Exp (Reasoning)OpenDeepSeek56.9$0.32——
39▼ 2Claude 4 Sonnet (Reasoning)ClosedAnthropic56.5$6.0——
40▼ 2Nova 2.0 Omni (medium)ClosedAmazon56.0$0.85——
41▼ 2GLM-4.6 (Reasoning)OpenZ AI56.0$1.0——
42▼ 2Qwen3 Max ThinkingClosedAlibaba55.8$2.4——
43▼ 2Qwen3 MaxClosedAlibaba55.2$2.4——
44▼ 2Claude 4.5 Haiku (Reasoning)ClosedAnthropic54.6$2.0——
45▼ 2Gemini 2.5 Flash Preview (Sep '25) (Reasoning)ClosedGoogle54.4$0.85——
46▼ 2Qwen3 VL 235B A22B (Reasoning)OpenAlibaba54.4$2.6——
47▼ 2Qwen3 Next 80B A3B (Reasoning)OpenAlibaba54.3$1.9——
48▼ 2Claude 4 Opus (Reasoning)ClosedAnthropic54.2$30——
49▼ 2Gemini 2.5 Pro Preview (Mar' 25)ClosedGoogle54.1$3.4——
50▼ 2DeepSeek V3.1 (Reasoning)OpenDeepSeek54.0$0.65——

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 “—”.