Free AI Models
FreeThe best zero-cost AI models ranked by intelligence score. All scores computed from real ANT routing data.
Updated October 13, 2026
| Rank | Model | Provider | Score | Success Rate | Avg Latency | Requests |
|---|---|---|---|---|---|---|
| 1 | OpenAI: GPT-5 Pro | openrouter | 97.0 | 0.0% | 2000ms | 0 |
| 2 | OpenAI: o3 Deep Research | openrouter | 92.0 | 0.0% | 1500ms | 0 |
| 3 | Grok 3 | xai | 88.0 | 0.0% | 400ms | 0 |
| 4 | OpenAI: GPT-5.1 | openrouter | 85.0 | 0.0% | 1000ms | 0 |
| 5 | OpenAI: GPT-5.1-Codex | openrouter | 85.0 | 0.0% | 1000ms | 0 |
| 6 | OpenAI: GPT-5 Codex | openrouter | 85.0 | 0.0% | 1000ms | 0 |
| 7 | OpenAI: GPT-5.1-Codex-Max | openrouter | 85.0 | 0.0% | 1000ms | 0 |
| 8 | OpenAI: o4 Mini Deep Research | openrouter | 85.0 | 0.0% | 1000ms | 0 |
| 9 | Rerank v3.5 | cohere | 80.0 | 0.0% | 400ms | 0 |
| 10 | OpenAI: GPT-5.4 Mini | openrouter | 79.0 | 0.0% | 600ms | 0 |
| 11 | OpenAI: GPT-4o Search Preview | openrouter | 79.0 | 0.0% | 1000ms | 0 |
| 12 | Llama 4 Maverick 17B 128E Instruct | huggingface | 78.0 | 0.0% | 400ms | 0 |
| 13 | command-a-03-2025 | semantic-cache | 77.0 | 10000.0% | 0ms | 14,828 |
| 14 | deepseek-ai/DeepSeek-V4.1-Flash | semantic-cache | 77.0 | 10000.0% | 0ms | 9,780 |
| 15 | gemini-2.5-flash | semantic-cache | 77.0 | 10000.0% | 0ms | 18,017 |
| 16 | meta-llama/Llama-4-Scout-17B-16E-Instruct | semantic-cache | 77.0 | 10000.0% | 0ms | 19,443 |
| 17 | Rerank Multilingual v3.0 | cohere | 75.0 | 0.0% | 400ms | 0 |
| 18 | Rerank English v3.0 | cohere | 75.0 | 0.0% | 400ms | 0 |
| 19 | Embed English v3.0 | cohere | 75.0 | 0.0% | 400ms | 0 |
| 20 | Embed Multilingual v3.0 | cohere | 75.0 | 0.0% | 400ms | 0 |
| 21 | OpenAI: GPT-5.1-Codex-Mini | openrouter | 75.0 | 0.0% | 600ms | 0 |
| 22 | OpenAI: GPT-5 Mini | openrouter | 75.0 | 0.0% | 600ms | 0 |
| 23 | Grok 3 Mini | xai | 72.0 | 0.0% | 400ms | 0 |
| 24 | QwQ 32B | huggingface | 72.0 | 0.0% | 400ms | 0 |
| 25 | OpenAI: GPT-3.5 Turbo 16k | openrouter | 72.0 | 0.0% | 1000ms | 0 |
| 26 | OpenAI: GPT-3.5 Turbo Instruct | openrouter | 68.0 | 0.0% | 1000ms | 0 |
| 27 | OpenAI: GPT-5 Nano | openrouter | 65.0 | 0.0% | 600ms | 0 |
| 28 | Llama 4 Maverick | groq | 64.0 | 10000.0% | 532ms | 1 |
| 29 | DeepSeek-V3-0324 | groq | 64.0 | 10000.0% | 494ms | 1 |
| 30 | Llama 4 Scout | huggingface | 63.0 | 10000.0% | 6664ms | 211 |
| 31 | Kimi K2.5 | novita | 63.0 | 10000.0% | 1085ms | 1 |
| 32 | Mistral Small 3.1 | huggingface | 62.0 | 10000.0% | 1657ms | 1 |
| 33 | deepseek-ai/DeepSeek-V3-0324 | semantic-cache | 62.0 | 10000.0% | 0ms | 191 |
| 34 | DeepSeek V3.1 | huggingface | 62.0 | 10000.0% | 5636ms | 261 |
| 35 | GPT-4o Mini (AskCodi) | huggingface | 62.0 | 10000.0% | 1883ms | 4 |
| 36 | anthropic/claude-opus-4.8@openrouter | openrouter | 61.0 | 10000.0% | 2400ms | 1 |
| 37 | deepseek/deepseek-r1-turbo | semantic-cache | 61.0 | 10000.0% | 0ms | 126 |
| 38 | Gpt 5.3 Codex | openrouter | 61.0 | 10000.0% | 2014ms | 1 |
| 39 | anthropic/claude-opus-4-6 | anthropic | 61.0 | 10000.0% | 2033ms | 1 |
| 40 | anthropic/claude-opus-4.8@vercel | vercel | 61.0 | 10000.0% | 2181ms | 2 |
| 41 | Mistral Large 2411 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 42 | Mistral Large Pixtral 2411 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 43 | Labs Mistral Small Creative | mistral | 60.0 | 0.0% | 400ms | 0 |
| 44 | Devstral Small 2507 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 45 | Embed Multilingual Light v3.0 | cohere | 60.0 | 0.0% | 400ms | 0 |
| 46 | Voxtral Mini Realtime Latest | mistral | 60.0 | 0.0% | 400ms | 0 |
| 47 | Mistral Small 2506 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 48 | Voxtral Mini 2507 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 49 | Devstral Medium 2507 | mistral | 60.0 | 0.0% | 400ms | 0 |
| 50 | Voxtral Small Latest | mistral | 60.0 | 0.0% | 400ms | 0 |