Skip to content
IFAI

Model rankings

669 models · 3 boards · 2 sources

Artificial AnalysisIntelligence IndexArenaText arena

Intelligence Index · A composite of several evaluations · Updated October 15, 2025

Smartest
GPT-5 Codex (high)OpenAI68.5Intelligence
Best open model
gpt-oss-120B (high)OpenAI58.0Intelligence
Best value
Gemma 3n E4B InstructGoogle620 pts/$Value
Best from China
DeepSeek V3.1 Terminus (Reasoning)DeepSeek57.7Intelligence
Top three by Intelligence
1GPT-5 Codex (high)OpenAI68.5
2GPT-5 (high)OpenAI68.5
3o3OpenAI65.5
Top three open models
1gpt-oss-120B (high)OpenAI · #12 overall58.0
2DeepSeek V3.1 Terminus (Reasoning)DeepSeek · #13 overall57.7
3Qwen3 235B A22B 2507 (Reasoning)Alibaba · #14 overall57.5
This week's changes
SRPOImage · #22New
Seed-OSS-36B-InstructIntelligence · #27New
ERNIE 4.5 300B A47BIntelligence · #82New
Jamba Reasoning 3BIntelligence · #133New
LFM2 8B A1BIntelligence · #152New
RankChangeModelIntelligencePrice $/MSpeed tok/sFirst token
1—gpt-oss-120B (high)OpenOpenAI58.0$0.26——
2—Apriel-v1.5-15B-ThinkerOpenServiceNow51.6$0.000——
3—Llama Nemotron Super 49B v1.5 (Reasoning)OpenNVIDIA · 1 more variant45.2$0.18——
4—gpt-oss-20B (high)OpenOpenAI43.3$0.095——
5—Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)OpenNVIDIA38.5$0.90——
6—NVIDIA Nemotron Nano 9B V2 (Reasoning)OpenNVIDIA · 1 more variant36.9$0.070——
7—Llama 4 MaverickOpenMeta35.8$0.40——
8—Llama 3.3 Nemotron Super 49B v1 (Reasoning)OpenNVIDIA · 1 more variant35.5$0.000——
9—Llama 4 ScoutOpenMeta28.1$0.24——
10▲ 2Llama 3.1 Instruct 405BOpenMeta28.1$3.4——
11▼ 1Llama 3.3 Instruct 70BOpenMeta27.9$0.54——
12▼ 1Llama 3.1 Nemotron Nano 4B v1.1 (Reasoning)OpenNVIDIA26.1$0.000——
13—Grok 2 (Dec '24)OpenxAI24.7$0.000——
14▲ 1Llama 3.1 Nemotron Instruct 70BOpenNVIDIA23.6$0.60——
15▲ 1Granite 4.0 H SmallOpenIBM22.7$0.11——
16▼ 2Phi-4OpenMicrosoft Azure22.7$0.22——
17—Llama 3.1 Instruct 70BOpenMeta22.6$0.72——
18—Gemma 3 27B InstructOpenGoogle22.2$0.000——
19—Gemma 3 12B InstructOpenGoogle20.4$0.000——
20—R1 1776OpenPerplexity19.1$3.5——
21—Llama 3.2 Instruct 90B (Vision)OpenMeta18.9$0.72——
22—Grok-1OpenxAI18.2$0.000——
23NewLFM2 8B A1BOpenLiquid AI17.4$0.000——
24▼ 1Gemma 2 27BOpenGoogle17.2$0.000——
25▼ 1Llama 3.1 Instruct 8BOpenMeta16.9$0.089——
26▼ 1Granite 4.0 MicroOpenIBM16.2$0.000——
27▼ 1Phi-4 Mini InstructOpenMicrosoft Azure15.7$0.000——
28▼ 1Llama 3.2 Instruct 11B (Vision)OpenMeta15.5$0.16——
29▼ 1Gemma 3n E4B InstructOpenGoogle15.5$0.025——
30▼ 1Granite 3.3 8B (Non-reasoning)OpenIBM15.2$0.085——
31▼ 1Gemma 3 4B InstructOpenGoogle14.7$0.000——
32▲ 6Phi-3 Medium Instruct 14BOpenMicrosoft Azure14.4$0.30——
33▼ 1Llama 3 Instruct 70BOpenMeta13.0$0.75——
34▼ 3Phi-3 Mini Instruct 3.8BOpenMicrosoft Azure12.7$0.23——
35▼ 2Gemma 3n E4B Instruct Preview (May '25)OpenGoogle12.5$0.000——
36▼ 2Phi-4 Multimodal InstructOpenMicrosoft Azure12.4$0.000——
37▼ 2LFM2 2.6BOpenLiquid AI11.8$0.000——
38▼ 2Llama 2 Chat 7BOpenMeta11.3$0.10——
39▲ 1Gemma 3n E2B InstructOpenGoogle11.3$0.000——
40▼ 3Llama 3.2 Instruct 3BOpenMeta11.2$0.048——
41▲ 2Llama 3.2 Instruct 1BOpenMeta8.9$0.053——
42▼ 3LFM2 1.2BOpenLiquid AI8.0$0.000——
43▼ 2Gemma 2 9BOpenGoogle7.8$0.045——
44▼ 2Arctic InstructOpenSnowflake7.6$0.000——
45▼ 1Llama 3 Instruct 8BOpenMeta7.0$0.055——
46▼ 1Gemma 3 1B InstructOpenGoogle6.8$0.000——
47NewGemma 3 270MOpenGoogle5.6$0.000——
48▼ 2Llama 2 Chat 70BOpenMeta5.6$0.000——
49▼ 2Llama 2 Chat 13BOpenMeta5.5$0.000——
50▼ 2DBRX InstructOpenDatabricks5.3$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 “—”.