Fastest AI models
The fastest AI models by output throughput (tokens generated per second) — useful for chat, agents and high-volume work. All available on anyAInow, pay-as-you-go.
| # | Model | Speed | |
|---|---|---|---|
| 1 | GPT-5.4 mini OpenAIFast·from $10 / 1M | 924 tok/s | Try → |
| 2 | Mistral Small 4 MistralFast·from $5 / 1M | 737 tok/s | Try → |
| 3 | Mistral Medium 3.5 MistralBalanced·from $20 / 1M | 730 tok/s | Try → |
| 4 | GPT-5 nano OpenAIFree·Free | 500 tok/s | Try → |
| 5 | GPT-5.4 OpenAIBalanced·from $30 / 1M | 467 tok/s | Try → |
| 6 | Gemini 2.5 Pro GoogleFlagship·from $20 / 1M | 394 tok/s | Try → |
| 7 | MiniMax M3 MiniMaxOpen·from $10 / 1M | 380 tok/s | Try → |
| 8 | Claude Haiku 4.5 AnthropicFast·from $10 / 1M | 346 tok/s | Try → |
| 9 | GPT-5.5 OpenAIFlagship·from $50 / 1M | 341 tok/s | Try → |
| 10 | Claude Sonnet 4.6 AnthropicBalanced·from $30 / 1M | 332 tok/s | Try → |
| 11 | Gemini 3.1 Pro GoogleFlagship·from $20 / 1M | 266 tok/s | Try → |
| 12 | Claude Opus 4.8 AnthropicFlagship·from $30 / 1M | 217 tok/s | Try → |
| 13 | GLM 5.2 Z.ai (GLM)Open·from $10 / 1M | 206 tok/s | Try → |
| 14 | Qwen 3.6 Plus QwenOpen·from $10 / 1M | 171 tok/s | Try → |
| 15 | Qwen 3.7 Plus QwenOpen·from $10 / 1M | 142 tok/s | Try → |
| 16 | Nova Lite AmazonFast·from $5 / 1M | 100 tok/s | Try → |
| 17 | Nova Pro AmazonBalanced·from $10 / 1M | 100 tok/s | Try → |
| 18 | Kimi K2 Moonshot (Kimi)Balanced·from $10 / 1M | 87 tok/s | Try → |
| 19 | Llama 4 Maverick Meta LlamaBalanced·from $10 / 1M | 84 tok/s | Try → |
| 20 | Llama 4 Scout Meta LlamaOpen·from $5 / 1M | 76 tok/s | Try → |
| 21 | DeepSeek R1 DeepSeekReasoning·from $10 / 1M | 45 tok/s | Try → |
| 22 | Llama 3.3 70B Meta LlamaFree·Free | 42 tok/s | Try → |
| 23 | Gemma 3 27B GoogleOpen·from $10 / 1M | 33 tok/s | Try → |
| 24 | Phi-4 MicrosoftOpen·from $10 / 1M | 33 tok/s | Try → |
Throughput via llm-stats.com (7-day average), as of 28 Jun 2026. Prices shown are the anyAInow pay-as-you-go floor (per 1M tokens); free-tier models deduct nothing.