Cheapest AI APIs: compare token prices
Compare reported API input and output prices per million tokens in USD. Sorted by input price, then output price. A low input price does not necessarily mean the lowest total bill: output and reasoning tokens matter too.
Which should you choose?
No single benchmark covers every use case. Try your own prompts and compare accuracy, response time, input/output support, and the actual bill.
AI models
| Model | Intelligence score | Benchmark cost (USD) | Input price / 1M tokens (USD) | Output price / 1M tokens (USD) | Output speed |
|---|---|---|---|---|---|
| Apriel-v1.5-15B-ThinkerServiceNow | 21.20 | - | 0.00 | 0.00 | - |
| Apriel-v1.6-15B-ThinkerServiceNow | 20.50 | - | 0.00 | 0.00 | - |
| Gemini 2.0 Flash (experimental)Google | 10.70 | - | 0.00 | 0.00 | - |
| Gemma 3 12B InstructGoogle | 5.50 | - | 0.00 | 0.00 | - |
| Gemma 3 1B InstructGoogle | 1.00 | - | 0.00 | 0.00 | - |
| Gemma 3 27B InstructGoogle | 7.40 | - | 0.00 | 0.00 | - |
| Gemma 3 4B InstructGoogle | 1.10 | - | 0.00 | 0.00 | - |
| Gemma 3n E2B InstructGoogle | 1.00 | - | 0.00 | 0.00 | - |
| Gemma 4 31B (Reasoning)Google | 29.40 | - | 0.00 | 0.00 | 36 tok/s |
| Hy3Tencent | 41.20 | - | 0.00 | 0.00 | 49 tok/s |
| LFM2.5-8B-A1BLiquid AI | 8.30 | - | 0.00 | 0.00 | 330 tok/s |
| LFM2.5-VL-1.6BLiquid AI | 1.00 | - | 0.00 | 0.00 | 390 tok/s |
| Olmo 3.1 32B ThinkAllen Institute for AI | 8.10 | - | 0.00 | 0.00 | - |
| Phi-4 Multimodal InstructMicrosoft | 4.50 | - | 0.00 | 0.00 | 18 tok/s |
| Sarvam M (Reasoning)Sarvam | 3.00 | - | 0.00 | 0.00 | - |
| Tiny Aya GlobalCohere | 1.00 | - | 0.00 | 0.00 | - |
| Gemma 3n E4B InstructGoogle | 1.20 | - | 0.02 | 0.04 | 53 tok/s |
| Sarvam 30B (high)Sarvam | 6.60 | - | 0.03 | 0.11 | 167 tok/s |
| Qwen3.5 4B (Non-reasoning)Alibaba | 16.00 | - | 0.03 | 0.15 | 18 tok/s |
| Qwen3.5 4B (Reasoning)Alibaba | 20.10 | - | 0.03 | 0.15 | 19 tok/s |
| Granite 3.3 8B (Non-reasoning)IBM | 1.80 | - | 0.03 | 0.25 | 346 tok/s |
| HyperNova 60B 2605Multiverse Computing | 17.80 | $35.0 | 0.04 | 0.14 | 357 tok/s |
| Llama 3 Instruct 8BMeta | 1.20 | - | 0.04 | 0.14 | - |
| Nova MicroAmazon | 4.70 | - | 0.04 | 0.14 | 288 tok/s |
| NVIDIA Nemotron Nano 9B V2 (Reasoning)NVIDIA | 8.80 | - | 0.04 | 0.16 | 77 tok/s |
| Sarvam 105B (high)Sarvam | 11.90 | - | 0.04 | 0.17 | 99 tok/s |
| Granite 4.1 8BIBM | 6.70 | - | 0.05 | 0.10 | 119 tok/s |
| gpt-oss-20b (high)OpenAI | 14.90 | $29.9 | 0.05 | 0.20 | 210 tok/s |
| NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)NVIDIA | 7.40 | - | 0.05 | 0.20 | 137 tok/s |
| NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)NVIDIA | 14.20 | $38.9 | 0.05 | 0.20 | 166 tok/s |
| NVIDIA Nemotron Nano 9B V2 (Non-reasoning)NVIDIA | 7.40 | - | 0.05 | 0.20 | 132 tok/s |
| Qwen2.5 TurboAlibaba | 6.30 | - | 0.05 | 0.20 | 106 tok/s |
| Llama 2 Chat 7BMeta | 4.30 | - | 0.05 | 0.25 | - |
| GPT-5 nano (high)OpenAI | 19.90 | - | 0.05 | 0.40 | 157 tok/s |
| GPT-5 nano (medium)OpenAI | 19.00 | - | 0.05 | 0.40 | 166 tok/s |
| GPT-5 nano (minimal)OpenAI | 8.00 | - | 0.05 | 0.40 | 165 tok/s |
| gpt-oss-20b (low)OpenAI | 14.30 | - | 0.06 | 0.20 | 236 tok/s |
| Nova LiteAmazon | 6.90 | - | 0.06 | 0.24 | 169 tok/s |
| Granite 4.0 H SmallIBM | 5.20 | - | 0.06 | 0.25 | 422 tok/s |
| Nemotron 3 Nano Omni 30B A3B ReasoningNVIDIA | 14.90 | - | 0.07 | 0.30 | 328 tok/s |
| GLM-4.7-Flash (Non-reasoning)Z AI | 15.50 | - | 0.07 | 0.40 | 139 tok/s |
| GLM-4.7-Flash (Reasoning)Z AI | 22.90 | - | 0.07 | 0.40 | 93 tok/s |
| Llama 3.1 Instruct 8BMeta | 7.60 | - | 0.10 | 0.10 | 147 tok/s |
| Ministral 3 3BMistral | 6.80 | $91.8 | 0.10 | 0.10 | 174 tok/s |
| Qwen3.5 9B (Reasoning)Alibaba | 21.40 | $179 | 0.10 | 0.15 | 59 tok/s |
| Apertus 8B InstructSwiss AI Initiative | 1.00 | - | 0.10 | 0.20 | - |
| Olmo 3 7B InstructAllen Institute for AI | 2.80 | - | 0.10 | 0.20 | - |
| Gemma 4 12B (Non-reasoning)Google | 13.20 | - | 0.10 | 0.30 | 103 tok/s |
| Gemma 4 12B (Reasoning)Google | 22.00 | - | 0.10 | 0.30 | 108 tok/s |
| Ling 2.6 FlashInclusionAI | 14.10 | - | 0.10 | 0.30 | 177 tok/s |
| MiMo-V2-Flash (Reasoning)Xiaomi | 31.20 | - | 0.10 | 0.30 | - |
| Mistral Small 3Mistral | 6.90 | - | 0.10 | 0.30 | 152 tok/s |
| Mistral Small 3.1Mistral | 14.70 | $29.9 | 0.10 | 0.30 | 152 tok/s |
| Mistral Small 3.2Mistral | 10.60 | $128 | 0.10 | 0.30 | 146 tok/s |
| Step 3.5 FlashStepFun | 25.50 | - | 0.10 | 0.30 | 281 tok/s |
| Step 3.5 Flash 2603StepFun | 26.00 | - | 0.10 | 0.30 | 268 tok/s |
| Gemini 2.5 Flash-Lite (Non-reasoning)Google | 6.90 | - | 0.10 | 0.40 | 203 tok/s |
| Gemini 2.5 Flash-Lite (Reasoning)Google | 11.40 | - | 0.10 | 0.40 | 255 tok/s |
| Gemini 2.5 Flash-Lite Preview (Sep '25) (Non-reasoning)Google | 13.10 | - | 0.10 | 0.40 | - |
| Gemini 2.5 Flash-Lite Preview (Sep '25) (Reasoning)Google | 15.10 | - | 0.10 | 0.40 | - |
| GPT-4.1 nanoOpenAI | 9.60 | $48.7 | 0.10 | 0.40 | 192 tok/s |
| Qwen3.5 Omni FlashAlibaba | 19.00 | - | 0.10 | 0.80 | 243 tok/s |
| Hy3-preview (Non-reasoning)Tencent | 26.10 | - | 0.12 | 0.43 | 139 tok/s |
| Hy3-preview (Reasoning)Tencent | 33.60 | - | 0.12 | 0.43 | 128 tok/s |
| Gemma 4 26B A4B (Non-reasoning)Google | 20.10 | - | 0.13 | 0.40 | 71 tok/s |
| Gemma 4 26B A4B (Reasoning)Google | 25.70 | $51.9 | 0.13 | 0.40 | - |
| Hermes 4 - Llama-3.1 70B (Non-reasoning)Nous Research | 6.90 | - | 0.13 | 0.40 | 82 tok/s |
| Hermes 4 - Llama-3.1 70B (Reasoning)Nous Research | 10.00 | - | 0.13 | 0.40 | 83 tok/s |
| Phi-4Microsoft | 4.90 | - | 0.13 | 0.50 | 36 tok/s |
| DeepSeek V4 Flash (Non-reasoning)DeepSeek | 28.70 | - | 0.14 | 0.28 | 117 tok/s |
| DeepSeek V4 Flash (Reasoning, High Effort)DeepSeek | 37.40 | - | 0.14 | 0.28 | - |
| DeepSeek V4 Flash (Reasoning, Max Effort)DeepSeekBest value frontier | 40.30 | $74.3 | 0.14 | 0.28 | 110 tok/s |
| MiMo-V2.5XiaomiBest value frontier | 37.20 | $25.2 | 0.14 | 0.28 | 64 tok/s |
| Gemma 4 31B (Non-reasoning)Google | 21.80 | $29.0 | 0.14 | 0.40 | 46 tok/s |
| Ling-flash-2.0InclusionAI | 9.70 | - | 0.14 | 0.57 | 73 tok/s |
| Ring-flash-2.0InclusionAI | 8.20 | - | 0.14 | 0.57 | - |
| Ministral 3 8BMistral | 9.00 | $113 | 0.15 | 0.15 | 97 tok/s |
| Solar MiniUpstage | 6.20 | - | 0.15 | 0.15 | - |
| Gemini 2.0 Flash (Feb '25)Google | 12.30 | - | 0.15 | 0.60 | - |
| GPT-4o miniOpenAI | 6.90 | - | 0.15 | 0.60 | 71 tok/s |
| gpt-oss-120b (high)OpenAI | 23.80 | $96.3 | 0.15 | 0.60 | 292 tok/s |
| gpt-oss-120b (low)OpenAIBest value frontier | 14.90 | $24.2 | 0.15 | 0.60 | 320 tok/s |
| Mistral Small 4 (Non-reasoning)Mistral | 12.40 | - | 0.15 | 0.60 | 157 tok/s |
| Mistral Small 4 (Reasoning)Mistral | 19.60 | - | 0.15 | 0.60 | 173 tok/s |
| Llama 4 ScoutMetaBest value frontier | 10.00 | $11.5 | 0.17 | 0.63 | 97 tok/s |
| GLM-4.5-AirZ AI | 16.50 | - | 0.17 | 0.98 | 75 tok/s |
| Qwen3 8B (Non-reasoning)Alibaba | 5.10 | - | 0.18 | 0.70 | 38 tok/s |
| Qwen3 VL 8B InstructAlibaba | 8.40 | - | 0.18 | 0.70 | 121 tok/s |
| Qwen3 8B (Reasoning)Alibaba | 8.30 | - | 0.18 | 2.10 | 38 tok/s |
| Qwen3 VL 8B (Reasoning)Alibaba | 10.60 | - | 0.18 | 2.10 | 115 tok/s |
| Ministral 3 14BMistral | 11.10 | $120 | 0.20 | 0.20 | 59 tok/s |
| Jamba 1.5 MiniAI21 Labs | 2.70 | - | 0.20 | 0.40 | - |
| Jamba 1.6 MiniAI21 Labs | 2.60 | - | 0.20 | 0.40 | 182 tok/s |
| Grok 4 Fast (Non-reasoning)SpaceXAI | 16.50 | - | 0.20 | 0.50 | - |
| Grok 4 Fast (Reasoning)SpaceXAI | 27.40 | - | 0.20 | 0.50 | - |
| Mistral Small (Sep '24)Mistral | 4.70 | - | 0.20 | 0.60 | 158 tok/s |
| NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)NVIDIA | 4.60 | - | 0.20 | 0.60 | 207 tok/s |
| NVIDIA Nemotron Nano 12B v2 VL (Reasoning)NVIDIA | 9.00 | - | 0.20 | 0.60 | 107 tok/s |
| Qwen3 30B A3B (Non-reasoning)Alibaba | 6.80 | - | 0.20 | 0.80 | 109 tok/s |
| Qwen3 30B A3B 2507 InstructAlibaba | 9.10 | - | 0.20 | 0.80 | 142 tok/s |
| Qwen3 VL 30B A3B InstructAlibaba | 10.00 | - | 0.20 | 0.80 | 111 tok/s |
| Reka Flash (Sep '24)Reka AI | 6.30 | - | 0.20 | 0.80 | 56 tok/s |
| Reka Flash 3Reka AI | 4.10 | - | 0.20 | 0.80 | - |
| Step 3.7 FlashStepFun | 30.30 | - | 0.20 | 1.15 | 385 tok/s |
| GPT-5.4 nano (medium)OpenAI | 30.20 | - | 0.20 | 1.25 | 163 tok/s |
| GPT-5.4 nano (Non-Reasoning)OpenAI | 17.60 | - | 0.20 | 1.25 | 171 tok/s |
| GPT-5.4 nano (xhigh)OpenAI | 38.20 | - | 0.20 | 1.25 | 161 tok/s |
| Qwen3 30B A3B (Reasoning)Alibaba | 9.30 | - | 0.20 | 2.40 | 103 tok/s |
| Qwen3 30B A3B 2507 (Reasoning)Alibaba | 14.40 | $152 | 0.20 | 2.40 | 149 tok/s |
| Qwen3 VL 30B A3B (Reasoning)Alibaba | 13.30 | - | 0.20 | 2.40 | 111 tok/s |
| Seed-OSS-36B-InstructByteDance Seed | 18.30 | - | 0.21 | 0.57 | 35 tok/s |
| Trinity Large ThinkingArcee AI | 24.50 | - | 0.23 | 0.88 | 204 tok/s |
| Mistral 7B InstructMistral | 2.10 | - | 0.25 | 0.25 | 102 tok/s |
| Mercury 2Inception | 25.30 | - | 0.25 | 0.75 | 782 tok/s |
| NVIDIA Nemotron 3 Super 120B A12B (Reasoning)NVIDIA | 25.40 | $254 | 0.25 | 0.78 | 147 tok/s |
| Qwen3 Omni 30B A3B (Reasoning)Alibaba | 9.60 | - | 0.25 | 0.97 | 100 tok/s |
| Qwen3 Omni 30B A3B InstructAlibaba | 5.10 | - | 0.25 | 0.97 | 97 tok/s |
| Claude 3 HaikuAnthropic | 3.90 | - | 0.25 | 1.25 | - |
| Qwen3.6 35B A3B (Reasoning)Alibaba | 31.60 | $334 | 0.25 | 1.49 | 150 tok/s |
| Gemini 3.1 Flash-LiteGoogle | 25.00 | $93.7 | 0.25 | 1.50 | 296 tok/s |
| GPT-5 mini (high)OpenAI | 25.30 | - | 0.25 | 2.00 | 91 tok/s |
| GPT-5 mini (medium)OpenAI | 30.90 | - | 0.25 | 2.00 | 87 tok/s |
| GPT-5 mini (minimal)OpenAI | 14.30 | - | 0.25 | 2.00 | 97 tok/s |
| GPT-5.1 Codex mini (high)OpenAI | 30.60 | - | 0.25 | 2.00 | 207 tok/s |
| Qwen3.5 35B A3B (Non-reasoning)Alibaba | 24.00 | $227 | 0.25 | 2.00 | 164 tok/s |
| Qwen3.5 35B A3B (Reasoning)Alibaba | 29.30 | - | 0.25 | 2.00 | 135 tok/s |
| DeepSeek V3.1 Terminus (Non-reasoning)DeepSeek | 21.40 | - | 0.27 | 1.00 | - |
| DeepSeek V3.2 (Non-reasoning)DeepSeek | 24.70 | - | 0.28 | 0.42 | - |
| DeepSeek V3.2 (Reasoning)DeepSeek | 32.00 | - | 0.28 | 0.42 | - |
| DeepSeek V3.2 Exp (Non-reasoning)DeepSeek | 21.30 | - | 0.28 | 0.42 | - |
| DeepSeek V3.2 Exp (Reasoning)DeepSeek | 25.40 | - | 0.28 | 0.42 | - |
| ERNIE 4.5 300B A47BBaidu | 9.00 | - | 0.28 | 1.10 | - |
| Grok 3 mini Reasoning (high)SpaceXAI | 22.50 | - | 0.30 | 0.50 | 69 tok/s |
| GLM-4.6V (Non-reasoning)Z AI | 11.00 | - | 0.30 | 0.90 | 62 tok/s |
| GLM-4.6V (Reasoning)Z AI | 16.80 | - | 0.30 | 0.90 | 49 tok/s |
| KAT Coder Pro V2KwaiKAT | 33.70 | - | 0.30 | 1.20 | - |
| MiniMax-M2MiniMax | 28.30 | - | 0.30 | 1.20 | 85 tok/s |
| MiniMax-M2.1MiniMax | 31.40 | - | 0.30 | 1.20 | 84 tok/s |
| MiniMax-M2.5MiniMax | 33.70 | - | 0.30 | 1.20 | 87 tok/s |
| MiniMax-M2.7MiniMax | 38.10 | - | 0.30 | 1.20 | 63 tok/s |
| MiniMax-M3MiniMaxBest value frontier | 44.40 | $204 | 0.30 | 1.20 | 85 tok/s |
| Qwen3.5 27B (Non-reasoning)Alibaba | 29.30 | - | 0.30 | 2.40 | 84 tok/s |
| Qwen3.5 27B (Reasoning)Alibaba | 33.80 | - | 0.30 | 2.40 | 78 tok/s |
| Gemini 2.5 Flash (Non-reasoning)Google | 14.10 | - | 0.30 | 2.50 | 219 tok/s |
| Gemini 2.5 Flash (Reasoning)Google | 20.10 | - | 0.30 | 2.50 | 231 tok/s |
| Ling-2.6-1TInclusionAI | 26.10 | - | 0.30 | 2.50 | - |
| Nova 2.0 Lite (high)Amazon | 18.20 | $635 | 0.30 | 2.50 | 165 tok/s |
| Nova 2.0 Lite (low)Amazon | 17.80 | - | 0.30 | 2.50 | 164 tok/s |
| Nova 2.0 Lite (medium)Amazon | 19.00 | - | 0.30 | 2.50 | 163 tok/s |
| Nova 2.0 Lite (Non-reasoning)Amazon | 11.80 | - | 0.30 | 2.50 | 145 tok/s |
| Nova 2.0 Omni (low)Amazon | 16.60 | - | 0.30 | 2.50 | - |
| Nova 2.0 Omni (medium)Amazon | 20.90 | - | 0.30 | 2.50 | - |
| Nova 2.0 Omni (Non-reasoning)Amazon | 10.50 | - | 0.30 | 2.50 | - |
| Ring-2.6-1TInclusionAI | 30.60 | $460 | 0.30 | 2.50 | 134 tok/s |
| Llama 4 MaverickMeta | 14.30 | $38.2 | 0.35 | 0.85 | 113 tok/s |
| Qwen3 Coder NextAlibaba | 21.10 | $247 | 0.35 | 1.20 | 98 tok/s |
| Qwen3 14B (Non-reasoning)Alibaba | 7.00 | - | 0.35 | 1.40 | 64 tok/s |
| Qwen3 14B (Reasoning)Alibaba | 10.40 | - | 0.35 | 4.20 | 63 tok/s |
| Llama 3.2 Instruct 11B (Vision)Meta | 3.30 | - | 0.36 | 0.36 | 52 tok/s |
| DeepSeek V3 (Dec '24)DeepSeekBest value frontier | 14.20 | $19.8 | 0.36 | 0.89 | - |
| Qwen3.6 35B A3B (Non-reasoning)Alibaba | 24.20 | $496 | 0.38 | 2.25 | 179 tok/s |
| Llama Nemotron Super 49B v1.5 (Non-reasoning)NVIDIA | 8.70 | - | 0.40 | 0.40 | 51 tok/s |
| Llama Nemotron Super 49B v1.5 (Reasoning)NVIDIA | 12.40 | - | 0.40 | 0.40 | 51 tok/s |
| GPT-4.1 miniOpenAI | 14.80 | $74.7 | 0.40 | 1.60 | 80 tok/s |
| Qwen3.7 PlusAlibaba | 39.00 | $303 | 0.40 | 1.60 | 53 tok/s |
| Mistral Medium 3Mistral | 12.50 | - | 0.40 | 2.00 | 47 tok/s |
| Mistral Medium 3.1Mistral | 14.70 | $132 | 0.40 | 2.00 | 91 tok/s |
| Qwen3.5 122B A10B (Non-reasoning)Alibaba | 27.60 | $216 | 0.40 | 3.20 | 144 tok/s |
| Qwen3.5 122B A10B (Reasoning)Alibaba | 32.30 | $447 | 0.40 | 3.20 | 135 tok/s |
| Qwen3.5 Omni PlusAlibaba | 30.60 | - | 0.40 | 4.80 | 55 tok/s |
| DeepSeek V4 Pro (Non-reasoning)DeepSeek | 31.20 | - | 0.43 | 0.87 | 67 tok/s |
| DeepSeek V4 Pro (Reasoning, High Effort)DeepSeek | 40.80 | - | 0.43 | 0.87 | 62 tok/s |
| DeepSeek V4 Pro (Reasoning, Max Effort)DeepSeekBest value frontier | 44.30 | $176 | 0.43 | 0.87 | 62 tok/s |
| MiMo-V2.5-ProXiaomiBest value frontier | 42.20 | $98.5 | 0.43 | 0.87 | 62 tok/s |
| Mixtral 8x7B InstructMistral | 2.40 | - | 0.45 | 0.70 | - |
| Qwen3 Coder 30B A3B InstructAlibaba | 13.60 | - | 0.45 | 2.25 | 96 tok/s |
| Qwen2.5 Instruct 72BAlibaba | 9.60 | - | 0.47 | 0.49 | - |
| Command-R (Mar '24)Cohere | 2.10 | - | 0.50 | 1.50 | - |
| GPT-3.5 TurboOpenAI | 3.60 | - | 0.50 | 1.50 | - |
| Magistral Small 1.2Mistral | 11.30 | - | 0.50 | 1.50 | 81 tok/s |
| Mistral Large 3Mistral | 15.90 | $71.0 | 0.50 | 1.50 | 51 tok/s |
| Qwen3 Next 80B A3B InstructAlibaba | 13.70 | - | 0.50 | 2.00 | 183 tok/s |
| Nex-N2-ProNex AGI | 41.00 | - | 0.50 | 2.50 | 139 tok/s |
| Gemini 3 Flash Preview (Non-reasoning)Google | 27.40 | - | 0.50 | 3.00 | 168 tok/s |
| Gemini 3 Flash Preview (Reasoning)Google | 37.80 | - | 0.50 | 3.00 | 173 tok/s |
| Qwen3.6 PlusAlibaba | 39.60 | $526 | 0.50 | 3.00 | 53 tok/s |
| Qwen3 Next 80B A3B (Reasoning)Alibaba | 16.70 | $401 | 0.50 | 6.00 | 184 tok/s |
| GLM-4.6 (Reasoning)Z AI | 28.70 | $313 | 0.55 | 2.20 | 46 tok/s |
| MiniMax M1 80kMiniMax | 17.70 | - | 0.55 | 2.20 | - |
| Llama 3.1 Instruct 70BMeta | 6.80 | - | 0.56 | 0.56 | 41 tok/s |
| DeepSeek V3.1 (Non-reasoning)DeepSeek | 21.00 | - | 0.56 | 1.68 | - |
| GLM-4.6 (Non-reasoning)Z AI | 23.00 | - | 0.57 | 2.20 | 67 tok/s |
| Kimi K2Kimi | 19.40 | - | 0.57 | 2.30 | 37 tok/s |
| Llama 3.3 Instruct 70BMeta | 9.40 | $78.0 | 0.58 | 0.71 | 86 tok/s |
| DeepSeek V3.1 (Reasoning)DeepSeek | 20.70 | - | 0.59 | 1.69 | - |
| GLM-4.5V (Non-reasoning)Z AI | 7.00 | - | 0.60 | 1.80 | 81 tok/s |
| GLM-4.5V (Reasoning)Z AI | 9.10 | - | 0.60 | 1.80 | 63 tok/s |
| Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)NVIDIA | 9.10 | - | 0.60 | 1.80 | 53 tok/s |
| GLM-4.7 (Non-reasoning)Z AI | 26.60 | - | 0.60 | 2.20 | 113 tok/s |
| GLM-4.7 (Reasoning)Z AI | 33.70 | $603 | 0.60 | 2.20 | 116 tok/s |
| Kimi K2 0905Kimi | 23.50 | - | 0.60 | 2.50 | 38 tok/s |
| Kimi K2 ThinkingKimi | 32.70 | - | 0.60 | 2.50 | 122 tok/s |
| Kimi K2.5 (Non-reasoning)Kimi | 29.40 | - | 0.60 | 3.00 | 50 tok/s |
| Kimi K2.5 (Reasoning)Kimi | 35.40 | - | 0.60 | 3.00 | 56 tok/s |
| Qwen3.5 397B A17B (Non-reasoning)Alibaba | 32.00 | - | 0.60 | 3.60 | 61 tok/s |
| Qwen3.5 397B A17B (Reasoning)Alibaba | 33.70 | $528 | 0.60 | 3.60 | 61 tok/s |
| Qwen3.6 27B (Non-reasoning)Alibaba | 30.50 | $356 | 0.60 | 3.60 | 57 tok/s |
| Qwen3.6 27B (Reasoning)Alibaba | 37.10 | $668 | 0.60 | 3.60 | 56 tok/s |
| Llama 3 Instruct 70BMeta | 3.50 | - | 0.65 | 2.75 | - |
| QwQ 32BAlibaba | 13.40 | - | 0.66 | 1.00 | 30 tok/s |
| MiMo-V2.5-Pro (Non-reasoning)Xiaomi | 27.90 | - | 0.66 | 1.72 | 61 tok/s |
| Nemotron 3 Ultra 550B A55B (Reasoning)NVIDIA | 37.80 | $435 | 0.68 | 2.67 | 189 tok/s |
| Hermes 3 - Llama-3.1 70BNous Research | 5.10 | - | 0.70 | 0.70 | 33 tok/s |
| DeepSeek R1 Distill Llama 70BDeepSeek | 9.90 | - | 0.70 | 1.05 | 24 tok/s |
| Qwen3 235B A22B (Non-reasoning)Alibaba | 10.90 | - | 0.70 | 2.80 | 58 tok/s |
| Qwen3 235B A22B 2507 InstructAlibaba | 18.20 | - | 0.70 | 2.80 | 59 tok/s |
| Qwen3 32B (Non-reasoning)Alibaba | 8.60 | - | 0.70 | 2.80 | 93 tok/s |
| Qwen3 VL 235B A22B InstructAlibaba | 14.30 | - | 0.70 | 2.80 | 51 tok/s |
| Qwen3 VL 32B InstructAlibaba | 11.10 | - | 0.70 | 2.80 | 70 tok/s |
| Qwen3 235B A22B (Reasoning)Alibaba | 13.40 | - | 0.70 | 8.40 | 59 tok/s |
| Qwen3 235B A22B 2507 (Reasoning)Alibaba | 19.60 | - | 0.70 | 8.40 | 56 tok/s |
| Qwen3 32B (Reasoning)Alibaba | 11.50 | - | 0.70 | 8.40 | 91 tok/s |
| Qwen3 VL 235B A22B (Reasoning)Alibaba | 20.60 | - | 0.70 | 8.40 | 56 tok/s |
| Qwen3 VL 32B (Reasoning)Alibaba | 17.90 | - | 0.70 | 8.40 | 89 tok/s |
| GPT-5.4 mini (medium)OpenAI | 29.80 | - | 0.75 | 4.50 | 172 tok/s |
| GPT-5.4 mini (Non-Reasoning)OpenAI | 16.60 | - | 0.75 | 4.50 | 177 tok/s |
| GPT-5.4 mini (xhigh)OpenAI | 40.00 | $1095 | 0.75 | 4.50 | 170 tok/s |
| Nova ProAmazon | 7.70 | - | 0.80 | 3.20 | - |
| Claude 3.5 HaikuAnthropic | 12.30 | - | 0.80 | 4.00 | - |
| Apertus 70B InstructSwiss AI Initiative | 2.40 | - | 0.82 | 2.92 | - |
| Kimi K2.6Kimi | 44.20 | $841 | 0.95 | 4.00 | 46 tok/s |
| Kimi K2.6 (Non-reasoning)Kimi | 34.60 | - | 0.95 | 4.00 | 46 tok/s |
| Kimi K2.7 CodeKimi | 41.90 | - | 0.95 | 4.00 | 49 tok/s |
| Grok Build 0.1 0616SpaceXAI | 39.80 | $375 | 1.00 | 2.00 | 102 tok/s |
| Hermes 4 - Llama-3.1 405B (Non-reasoning)Nous Research | 8.80 | - | 1.00 | 3.00 | 39 tok/s |
| Hermes 4 - Llama-3.1 405B (Reasoning)Nous Research | 9.00 | - | 1.00 | 3.00 | 38 tok/s |
| Mistral Small (Feb '24)Mistral | 3.60 | - | 1.00 | 3.00 | 150 tok/s |
| GLM-5 (Non-reasoning)Z AI | 32.40 | - | 1.00 | 3.20 | 55 tok/s |
| GLM-5 (Reasoning)Z AI | 39.50 | - | 1.00 | 3.20 | 52 tok/s |
| Claude 4.5 Haiku (Non-reasoning)Anthropic | 23.70 | - | 1.00 | 5.00 | 89 tok/s |
| Claude 4.5 Haiku (Reasoning)Anthropic | 29.60 | $564 | 1.00 | 5.00 | 90 tok/s |
| GPT-5.6 Luna (high)OpenAIBest value frontier | 46.10 | $275 | 1.00 | 6.00 | 179 tok/s |
| GPT-5.6 Luna (low)OpenAI | 33.30 | $68.8 | 1.00 | 6.00 | 172 tok/s |
| GPT-5.6 Luna (max)OpenAI | 51.20 | $870 | 1.00 | 6.00 | 187 tok/s |
| GPT-5.6 Luna (medium)OpenAI | 38.10 | $106 | 1.00 | 6.00 | 177 tok/s |
| GPT-5.6 Luna (Non-reasoning)OpenAI | 26.60 | $51.4 | 1.00 | 6.00 | 183 tok/s |
| GPT-5.6 Luna (xhigh)OpenAI | 49.10 | $479 | 1.00 | 6.00 | 176 tok/s |
| o3-miniOpenAI | 19.00 | - | 1.10 | 4.40 | 215 tok/s |
| o3-mini (high)OpenAI | 15.60 | - | 1.10 | 4.40 | 220 tok/s |
| o4-mini (high)OpenAI | 25.60 | - | 1.10 | 4.40 | 172 tok/s |
| DeepSeek V3 0324DeepSeek | 15.40 | $132 | 1.14 | 1.25 | - |
| Llama 3.1 Nemotron Instruct 70BNVIDIA | 7.60 | - | 1.20 | 1.20 | 100 tok/s |
| Qwen3 MaxAlibaba | 24.00 | - | 1.20 | 6.00 | 55 tok/s |
| Qwen3 Max (Preview)Alibaba | 19.20 | - | 1.20 | 6.00 | 61 tok/s |
| Qwen3 Max Thinking (Preview)Alibaba | 25.00 | - | 1.20 | 6.00 | 51 tok/s |
| Cogito v2.1 (Reasoning)Deep Cogito | - | - | 1.25 | 1.25 | - |
| Grok 4.3 (high)SpaceXAI | 37.60 | $288 | 1.25 | 2.50 | 129 tok/s |
| Grok 4.3 (low)SpaceXAI | 35.40 | - | 1.25 | 2.50 | 109 tok/s |
| Grok 4.3 (medium)SpaceXAI | 36.00 | - | 1.25 | 2.50 | 123 tok/s |
| Grok 4.3 (Non-reasoning)SpaceXAI | 24.80 | $201 | 1.25 | 2.50 | 113 tok/s |
| Muse Spark 1.1 (xhigh)MetaBest value frontier | 50.60 | $548 | 1.25 | 4.25 | 115 tok/s |
| Gemini 2.5 ProGoogle | 25.80 | $610 | 1.25 | 10.00 | 143 tok/s |
| Gemini 2.5 Pro Preview (May' 25)Google | 22.30 | - | 1.25 | 10.00 | - |
| GPT-5 (ChatGPT)OpenAI | 15.30 | - | 1.25 | 10.00 | 166 tok/s |
| GPT-5 (high)OpenAI | 34.70 | $813 | 1.25 | 10.00 | 96 tok/s |
| GPT-5 (low)OpenAI | 31.20 | - | 1.25 | 10.00 | 75 tok/s |
| GPT-5 (medium)OpenAI | 33.70 | - | 1.25 | 10.00 | 91 tok/s |
| GPT-5 (minimal)OpenAI | 17.20 | - | 1.25 | 10.00 | 76 tok/s |
| GPT-5 Codex (high)OpenAI | 36.10 | - | 1.25 | 10.00 | 166 tok/s |
| GPT-5.1 (high)OpenAI | 36.90 | - | 1.25 | 10.00 | 100 tok/s |
| GPT-5.1 (Non-reasoning)OpenAI | 20.40 | - | 1.25 | 10.00 | 91 tok/s |
| GPT-5.1 Codex (high)OpenAI | 34.70 | - | 1.25 | 10.00 | 172 tok/s |
| Nova 2.0 Pro Preview (low)Amazon | 19.60 | $244 | 1.25 | 10.00 | 127 tok/s |
| Nova 2.0 Pro Preview (medium)Amazon | 21.80 | $407 | 1.25 | 10.00 | 129 tok/s |
| Nova 2.0 Pro Preview (Non-reasoning)Amazon | 14.40 | $206 | 1.25 | 10.00 | 110 tok/s |
| Qwen3.6 Max PreviewAlibaba | 40.00 | - | 1.30 | 7.80 | 40 tok/s |
| DeepSeek R1 0528 (May '25)DeepSeek | 20.10 | - | 1.35 | 4.20 | - |
| GLM-5.1 (Non-reasoning)Z AI | 35.40 | - | 1.40 | 4.40 | 56 tok/s |
| GLM-5.1 (Reasoning)Z AI | 40.20 | $659 | 1.40 | 4.40 | 68 tok/s |
| GLM-5.2 (max)Z AI | 51.10 | $925 | 1.40 | 4.40 | 182 tok/s |
| Mistral Medium 3.5Mistral | 29.90 | $1483 | 1.50 | 7.50 | 57 tok/s |
| Qwen3 Coder 480B A35B InstructAlibaba | 18.00 | - | 1.50 | 7.50 | 58 tok/s |
| Gemini 3.5 Flash (high)Google | 50.20 | $1041 | 1.50 | 9.00 | 161 tok/s |
| Gemini 3.5 Flash (medium)Google | 45.40 | - | 1.50 | 9.00 | 162 tok/s |
| Gemini 3.5 Flash (minimal)Google | 34.90 | - | 1.50 | 9.00 | 145 tok/s |
| DeepSeek V3.1 Terminus (Reasoning)DeepSeek | 30.40 | - | 1.64 | 2.75 | - |
| DeepSeek R1 (Jan '25)DeepSeek | 18.50 | $416 | 1.68 | 4.70 | - |
| GPT-5.2 (medium)OpenAI | 38.00 | - | 1.75 | 14.00 | - |
| GPT-5.2 (Non-reasoning)OpenAI | 26.00 | - | 1.75 | 14.00 | 66 tok/s |
| GPT-5.2 (xhigh)OpenAI | 42.20 | - | 1.75 | 14.00 | 74 tok/s |
| GPT-5.2 Codex (xhigh)OpenAI | 40.10 | - | 1.75 | 14.00 | 174 tok/s |
| GPT-5.3 Codex (xhigh)OpenAI | 44.30 | - | 1.75 | 14.00 | 86 tok/s |
| Inkling (xhigh)Thinking Machines | 40.70 | - | 1.87 | 4.68 | 85 tok/s |
| Magistral Medium 1.2Mistral | 17.90 | $886 | 2.00 | 5.00 | 44 tok/s |
| Grok 4.20 0309 (Non-reasoning)SpaceXAI | 22.50 | - | 2.00 | 6.00 | - |
| Grok 4.20 0309 (Reasoning)SpaceXAI | 36.50 | - | 2.00 | 6.00 | - |
| Grok 4.20 0309 v2 (Non-reasoning)SpaceXAI | 21.80 | - | 2.00 | 6.00 | 162 tok/s |
| Grok 4.20 0309 v2 (Reasoning)SpaceXAI | 37.00 | - | 2.00 | 6.00 | 176 tok/s |
| Grok 4.5 (high)SpaceXAIBest value frontier | 53.80 | $602 | 2.00 | 6.00 | 84 tok/s |
| Mistral Large 2 (Jul '24)Mistral | 7.30 | - | 2.00 | 6.00 | - |
| GPT-4.1OpenAI | 19.40 | - | 2.00 | 8.00 | 122 tok/s |
| Jamba 1.5 LargeAI21 Labs | 5.10 | - | 2.00 | 8.00 | - |
| Jamba 1.6 LargeAI21 Labs | 5.00 | - | 2.00 | 8.00 | 56 tok/s |
| Jamba 1.7 LargeAI21 Labs | 5.30 | - | 2.00 | 8.00 | 57 tok/s |
| o3OpenAI | 30.40 | - | 2.00 | 8.00 | 122 tok/s |
| Claude Sonnet 5 (Adaptive Reasoning, High Effort)Anthropic | - | - | 2.00 | 10.00 | 66 tok/s |
| Claude Sonnet 5 (Adaptive Reasoning, Low Effort)Anthropic | - | - | 2.00 | 10.00 | 60 tok/s |
| Claude Sonnet 5 (Adaptive Reasoning, Max Effort)Anthropic | 53.40 | $4010 | 2.00 | 10.00 | 83 tok/s |
| Claude Sonnet 5 (Adaptive Reasoning, Medium Effort)Anthropic | - | - | 2.00 | 10.00 | 64 tok/s |
| Claude Sonnet 5 (Adaptive Reasoning, Xhigh Effort)Anthropic | - | - | 2.00 | 10.00 | 70 tok/s |
| Claude Sonnet 5 (Non-reasoning, High Effort)Anthropic | 41.70 | $427 | 2.00 | 10.00 | 65 tok/s |
| Gemini 3 Pro Preview (high)Google | 39.60 | - | 2.00 | 12.00 | - |
| Gemini 3 Pro Preview (low)Google | 33.10 | - | 2.00 | 12.00 | - |
| Gemini 3.1 Pro PreviewGoogle | 46.50 | $815 | 2.00 | 12.00 | 116 tok/s |
| Llama 3.2 Instruct 90B (Vision)Meta | 6.20 | - | 2.04 | 2.04 | - |
| GLM-5.2 (Non-reasoning)Z AI | 34.10 | - | 2.09 | 6.38 | 75 tok/s |
| Qwen3.7 MaxAlibaba | 46.00 | $1559 | 2.50 | 7.50 | 202 tok/s |
| Command ACohere | 7.70 | - | 2.50 | 10.00 | 56 tok/s |
| GPT-4o (Aug '24)OpenAI | 9.60 | - | 2.50 | 10.00 | 84 tok/s |
| GPT-4o (Nov '24)OpenAI | 11.20 | - | 2.50 | 10.00 | 186 tok/s |
| Nova PremierAmazon | 12.70 | - | 2.50 | 12.50 | 28 tok/s |
| GPT-5.4 (low)OpenAI | 39.10 | - | 2.50 | 15.00 | 107 tok/s |
| GPT-5.4 (Non-reasoning)OpenAI | 27.70 | - | 2.50 | 15.00 | 107 tok/s |
| GPT-5.4 (xhigh)OpenAI | 51.40 | $2132 | 2.50 | 15.00 | 136 tok/s |
| GPT-5.6 Terra (high)OpenAI | 49.00 | $496 | 2.50 | 15.00 | 120 tok/s |
| GPT-5.6 Terra (low)OpenAI | 40.50 | $161 | 2.50 | 15.00 | 120 tok/s |
| GPT-5.6 Terra (max)OpenAI | 55.00 | $1754 | 2.50 | 15.00 | 136 tok/s |
| GPT-5.6 Terra (medium)OpenAIBest value frontier | 45.60 | $240 | 2.50 | 15.00 | 125 tok/s |
| GPT-5.6 Terra (Non-reasoning)OpenAI | 34.00 | - | 2.50 | 15.00 | 123 tok/s |
| GPT-5.6 Terra (xhigh)OpenAI | 51.60 | $740 | 2.50 | 15.00 | 123 tok/s |
| Mistral MediumMistral | 3.60 | - | 2.75 | 8.10 | 74 tok/s |
| Claude 3 SonnetAnthropic | 4.70 | - | 3.00 | 15.00 | - |
| Claude 3.5 Sonnet (June '24)Anthropic | 8.30 | - | 3.00 | 15.00 | - |
| Claude 3.5 Sonnet (Oct '24)Anthropic | 9.90 | - | 3.00 | 15.00 | - |
| Claude 3.7 Sonnet (Non-reasoning)Anthropic | 23.50 | - | 3.00 | 15.00 | - |
| Claude 4 Sonnet (Non-reasoning)Anthropic | 25.50 | - | 3.00 | 15.00 | - |
| Claude 4 Sonnet (Reasoning)Anthropic | 28.90 | - | 3.00 | 15.00 | - |
| Claude 4.5 Sonnet (Non-reasoning)Anthropic | 29.30 | - | 3.00 | 15.00 | 40 tok/s |
| Claude 4.5 Sonnet (Reasoning)Anthropic | 36.40 | $1133 | 3.00 | 15.00 | 43 tok/s |
| Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)Anthropic | 47.20 | $3356 | 3.00 | 15.00 | 51 tok/s |
| Claude Sonnet 4.6 (Non-reasoning, High Effort)Anthropic | 35.90 | - | 3.00 | 15.00 | 43 tok/s |
| Claude Sonnet 4.6 (Non-reasoning, Low Effort)Anthropic | 34.30 | - | 3.00 | 15.00 | 43 tok/s |
| Command-R+ (Apr '24)Cohere | 3.00 | - | 3.00 | 15.00 | - |
| Kimi K3Kimi | 57.10 | $2710 | 3.00 | 15.00 | 62 tok/s |
| Llama 3.1 Instruct 405BMeta | 8.50 | - | 3.92 | 13.00 | - |
| Mistral Large (Feb '24)Mistral | 4.40 | - | 4.00 | 12.00 | - |
| Grok 3SpaceXAI | 18.40 | - | 4.00 | 20.00 | - |
| GPT-4o (May '24)OpenAI | 8.60 | - | 5.00 | 15.00 | 88 tok/s |
| Claude Opus 4.5 (Non-reasoning)Anthropic | 34.70 | - | 5.00 | 25.00 | 44 tok/s |
| Claude Opus 4.5 (Reasoning)Anthropic | 40.80 | - | 5.00 | 25.00 | 46 tok/s |
| Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Anthropic | 43.70 | - | 5.00 | 25.00 | 41 tok/s |
| Claude Opus 4.6 (Non-reasoning, High Effort)Anthropic | 37.80 | - | 5.00 | 25.00 | 38 tok/s |
| Claude Opus 4.7 (Adaptive Reasoning, Max Effort)Anthropic | 53.50 | $3738 | 5.00 | 25.00 | 53 tok/s |
| Claude Opus 4.7 (Non-reasoning, High Effort)Anthropic | 42.70 | - | 5.00 | 25.00 | 47 tok/s |
| Claude Opus 4.8 (Adaptive Reasoning, Max Effort)Anthropic | 55.70 | $3753 | 5.00 | 25.00 | 58 tok/s |
| GPT-5.5 (high)OpenAI | 53.10 | $1719 | 5.00 | 30.00 | 71 tok/s |
| GPT-5.5 (low)OpenAI | 43.50 | $375 | 5.00 | 30.00 | 69 tok/s |
| GPT-5.5 (medium)OpenAI | 50.40 | $936 | 5.00 | 30.00 | 72 tok/s |
| GPT-5.5 (Non-reasoning)OpenAI | 35.40 | $211 | 5.00 | 30.00 | 70 tok/s |
| GPT-5.5 (xhigh)OpenAI | 54.80 | $2778 | 5.00 | 30.00 | 69 tok/s |
| GPT-5.5 Instant (June 2026)OpenAI | 28.90 | $528 | 5.00 | 30.00 | - |
| GPT-5.5 Instant (May 2026)OpenAI | 33.50 | - | 5.00 | 30.00 | - |
| GPT-5.6 Sol (high)OpenAIBest value frontier | 55.90 | $956 | 5.00 | 30.00 | 49 tok/s |
| GPT-5.6 Sol (low)OpenAIBest value frontier | 49.40 | $353 | 5.00 | 30.00 | 51 tok/s |
| GPT-5.6 Sol (max)OpenAIBest value frontier | 58.90 | $2824 | 5.00 | 30.00 | 55 tok/s |
| GPT-5.6 Sol (medium)OpenAIBest value frontier | 53.60 | $593 | 5.00 | 30.00 | 51 tok/s |
| GPT-5.6 Sol (Non-reasoning)OpenAI | 41.20 | $242 | 5.00 | 30.00 | 58 tok/s |
| GPT-5.6 Sol (xhigh)OpenAIBest value frontier | 57.70 | $1543 | 5.00 | 30.00 | 54 tok/s |
| Grok 4SpaceXAI | 33.30 | - | 5.50 | 27.50 | - |
| GPT-4 TurboOpenAI | 7.90 | - | 10.00 | 30.00 | 32 tok/s |
| Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)AnthropicBest value frontier | 59.90 | $5631 | 10.00 | 50.00 | 61 tok/s |
| o1OpenAI | 23.40 | - | 15.00 | 60.00 | 125 tok/s |
| Claude 3 OpusAnthropic | 11.80 | - | 15.00 | 75.00 | - |
| Claude 4 Opus (Non-reasoning)Anthropic | 25.50 | - | 15.00 | 75.00 | - |
| Claude 4 Opus (Reasoning)Anthropic | 31.00 | - | 15.00 | 75.00 | - |
| Claude 4.1 Opus (Non-reasoning)Anthropic | 28.20 | - | 15.00 | 75.00 | 28 tok/s |
| Claude 4.1 Opus (Reasoning)Anthropic | 33.70 | - | 15.00 | 75.00 | 31 tok/s |
| o1-previewOpenAI | 17.00 | - | 16.50 | 66.00 | - |
| o3-proOpenAI | 32.50 | - | 20.00 | 80.00 | 37 tok/s |
| GPT-4OpenAI | 7.00 | - | 30.00 | 60.00 | - |
| GPT-5.4 Pro (xhigh)OpenAI | - | - | 30.00 | 180 | - |
| o1-proOpenAI | 18.90 | - | 150 | 600 | - |
Guides
How we compare models
Intelligence uses the Artificial Analysis Intelligence Index. Benchmark cost is the total USD cost of running that test, including reasoning tokens; it is not a per-token API price. Rankings use the same dataset and calculations in every language. General benchmark scores do not measure performance in a particular language or task.