Model rankings
326 models · 2 boards · 2 sources
Smartest
Best open model
Best from China
Top three by Intelligence
o4-mini (high)OpenAI69.8
Gemini 2.5 Pro Preview (Mar' 25)Google67.8
o3OpenAI66.8
Top three open models
Qwen3 235B A22B (Reasoning)Alibaba · #6 overall62.3
Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA · #8 overall60.8
DeepSeek R1DeepSeek · #10 overall60.2
This week's changes
Qwen3 235B A22B (Reasoning)Intelligence · #6New
Qwen3 32B (Reasoning)Intelligence · #11New
Qwen3 14B (Reasoning)Intelligence · #14New
Qwen3 30B A3B (Reasoning)Intelligence · #15New
Nova PremierIntelligence · #37New
| Rank | Change | Model | Intelligence | Price $/M | Speed tok/s | First token |
|---|---|---|---|---|---|---|
| 1 | — | o4-mini (high)ClosedOpenAI | 69.8 | — | — | — |
| 2 | — | Gemini 2.5 Pro Preview (Mar' 25)ClosedGoogle | 67.8 | — | — | — |
| 3 | — | o3ClosedOpenAI | 66.8 | — | — | — |
| 4 | — | Grok 3 mini Reasoning (high)ClosedxAI | 66.7 | — | — | — |
| 5 | — | o3-mini (high)ClosedOpenAI · 1 more variant | 65.5 | — | — | — |
| 6 | New | Qwen3 235B A22B (Reasoning)OpenAlibaba | 62.3 | — | — | — |
| 7 | ▼ 1 | o1ClosedOpenAI | 61.9 | — | — | — |
| 8 | ▼ 1 | Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)OpenNVIDIA | 60.8 | — | — | — |
| 9 | ▼ 1 | Gemini 2.5 Flash Preview (Reasoning)ClosedGoogle · 1 more variant | 60.5 | — | — | — |
| 10 | ▼ 1 | DeepSeek R1OpenDeepSeek | 60.2 | — | — | — |
| 11 | New | Qwen3 32B (Reasoning)OpenAlibaba | 59.2 | — | — | — |
| 12 | ▼ 2 | QwQ 32BOpenAlibaba | 58.1 | — | — | — |
| 13 | ▼ 2 | Claude 3.7 Sonnet (Extended Thinking)ClosedAnthropic · 1 more variant | 57.4 | — | — | — |
| 14 | New | Qwen3 14B (Reasoning)OpenAlibaba | 55.7 | — | — | — |
| 15 | New | Qwen3 30B A3B (Reasoning)OpenAlibaba | 55.6 | — | — | — |
| 16 | ▼ 4 | o1-miniClosedOpenAI | 53.8 | — | — | — |
| 17 | ▼ 4 | DeepSeek V3 0324 (Mar' 25)OpenDeepSeek · 1 more variant | 53.2 | — | — | — |
| 18 | ▼ 4 | GPT-4.1 miniClosedOpenAI | 52.9 | — | — | — |
| 19 | ▼ 4 | GPT-4.1ClosedOpenAI | 52.6 | — | — | — |
| 20 | ▼ 4 | Gemini 2.0 Flash Thinking Experimental (Jan '25)ClosedGoogle | 52.3 | — | — | — |
| 21 | ▼ 4 | DeepSeek R1 Distill Qwen 32BOpenDeepSeek | 51.9 | — | — | — |
| 22 | ▼ 4 | Grok 3ClosedxAI | 50.6 | — | — | — |
| 23 | ▼ 4 | Llama 4 MaverickOpenMeta | 50.5 | — | — | — |
| 24 | ▼ 4 | GPT-4o (March 2025, chatgpt-4o-latest)ClosedOpenAI · 2 more variants | 50.3 | — | — | — |
| 25 | ▼ 4 | Gemini 2.0 Pro Experimental (Feb '25)ClosedGoogle | 49.2 | — | — | — |
| 26 | ▼ 4 | DeepSeek R1 Distill Qwen 14BOpenDeepSeek | 49.0 | — | — | — |
| 27 | ▼ 4 | DeepSeek R1 Distill Llama 70BOpenDeepSeek | 48.2 | — | — | — |
| 28 | ▼ 4 | Gemini 2.0 Flash (Feb '25)ClosedGoogle · 1 more variant | 48.1 | — | — | — |
| 29 | ▼ 4 | Reka Flash 3OpenReka AI | 47.1 | — | — | — |
| 30 | ▼ 4 | Qwen2.5 MaxClosedAlibaba | 45.3 | — | — | — |
| 31 | ▼ 4 | Gemini 1.5 Pro (Sep '24)ClosedGoogle · 1 more variant | 44.6 | — | — | — |
| 32 | ▼ 4 | Claude 3.5 Sonnet (Oct '24)ClosedAnthropic | 44.4 | — | — | — |
| 33 | ▼ 4 | SonarClosedPerplexity | 43.4 | — | — | — |
| 34 | ▼ 4 | Llama 4 ScoutOpenMeta | 43.0 | — | — | — |
| 35 | ▼ 4 | Sonar ProClosedPerplexity | 42.7 | — | — | — |
| 36 | ▼ 4 | QwQ 32B-PreviewOpenAlibaba | 42.6 | — | — | — |
| 37 | New | Nova PremierClosedAmazon | 42.6 | — | — | — |
| 38 | ▼ 5 | Gemini 2.0 Flash-Lite (Feb '25)ClosedGoogle | 41.4 | — | — | — |
| 39 | ▼ 5 | Llama 3.3 Instruct 70BOpenMeta | 41.1 | — | — | — |
| 40 | ▼ 5 | GPT-4.1 nanoClosedOpenAI | 41.0 | — | — | — |
| 41 | ▼ 5 | Llama 3.1 Instruct 405BOpenMeta | 40.5 | — | — | — |
| 42 | ▼ 5 | Qwen2.5 Instruct 72BOpenAlibaba | 40.4 | — | — | — |
| 43 | ▼ 5 | MiniMax-Text-01OpenMiniMax | 40.2 | — | — | — |
| 44 | ▼ 5 | Phi-4OpenMicrosoft Azure | 40.2 | — | — | — |
| 45 | ▼ 5 | Command AOpenCohere | 40.0 | — | — | — |
| 46 | ▼ 5 | Llama 3.1 Tulu3 405BOpenAllen Institute for AI | 39.9 | — | — | — |
| 47 | ▼ 5 | Llama 3.3 Nemotron Super 49B v1OpenNVIDIA | 39.3 | — | — | — |
| 48 | ▼ 5 | Grok 2 (Dec '24)ClosedxAI | 39.2 | — | — | — |
| 49 | ▼ 5 | Gemini 1.5 Flash (Sep '24)ClosedGoogle · 1 more variant | 39.0 | — | — | — |
| 50 | ▼ 5 | Mistral Large 2 (Nov '24)OpenMistral · 1 more variant | 38.3 | — | — | — |
Intelligence over time
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