LLM API pricing
178 models · USD per 1M tokens
last refresh 3 Nov 2026, 06:24 UTC178 / 215
Tags
| Model | Blended ↑ | Developer | In | Out | Ctx | tok/s | |
|---|---|---|---|---|---|---|---|
| cheapestgpt-oss-20breasoningprompt cachetools | $0.01 | OpenAI · via PrimeIntellect | 0.01 | 0.03 | 131K | - | |
| Qwen3.6 35B A3Bopen weightsreasoningvision | $0.03 | Alibaba · via PrimeIntellect | 0.01 | 0.09 | 262K | - | |
| gpt-oss-120breasoningprompt cachetools | $0.04 | OpenAI · via PrimeIntellect | 0.02 | 0.09 | 131K | - | |
| GLM 5.3 Flashopen weightsreasoningvision | $0.04 | Zhipu · via Relace | 0.03 | 0.09 | 1M | - | |
| Qwen3.5-9Bopen weightsreasoningvision | $0.05 | Alibaba · via PrimeIntellect | 0.05 | 0.07 | 262K | - | |
| Qwen3.7 Flashopen weightsreasoningvision | $0.06 | Alibaba | 0.03 | 0.13 | 1M | - | |
| Mistral Small 3open weights | $0.06 | Mistral | 0.05 | 0.08 | 33K | - | |
| Command R7B (12-2024) | $0.07 | Cohere | 0.04 | 0.15 | 128K | - | |
| Phi 4open weights | $0.07 | Microsoft | 0.06 | 0.11 | 16K | - | |
| Gemma 4 26B A4Bopen weightsreasoningvision | $0.07 | Google · via Darkbloom | 0.04 | 0.15 | 262K | - | |
| GPT-5 Nanoreasoningvisionprompt cache | $0.07 | OpenAI · via OpenAI | 0.03 | 0.20 | 400K | - | |
| Granite 4.2 8Bopen weightsreasoningprompt cache | $0.08 | IBM · via CoreWeave | 0.06 | 0.15 | 131K | - | |
| Gemini 2.5 Flash Litereasoningvisionprompt cache | $0.09 | 0.05 | 0.20 | 1M | - | ||
| Qwen3.8 27Bopen weightsreasoningvision | $0.09 | Alibaba · via PrimeIntellect | 0.01 | 0.33 | 1M | - | |
| DeepSeek V4 Flash 0423open weightsreasoningprompt cache | $0.10 | DeepSeek · via StreamLake | 0.08 | 0.17 | 1M | - | |
| Gemma 3 27Bopen weightsvisionprompt cache | $0.10 | Google · via DeepInfra | 0.08 | 0.16 | 131K | - | |
| GPT-6 Lunareasoningvisionprompt cache | $0.10 | OpenAI · via OpenAI | 0.05 | 0.25 | 1M | - | |
| GPT-6 Luna Proreasoningvisionprompt cache | $0.10 | OpenAI · via OpenAI | 0.05 | 0.25 | 1M | - | |
| Nova Lite 1.0visiontools200k+ context | $0.11 | Amazon | 0.06 | 0.24 | 300K | - | |
| Mistral Small 3.2 24Bopen weightsvisionprompt cache | $0.11 | Mistral · via DeepInfra | 0.07 | 0.20 | 256K | - | |
| Qwen3.5-Flashopen weightsreasoningvision | $0.11 | Alibaba | 0.07 | 0.26 | 1M | - | |
| DeepSeek V4.1 Flashopen weightsreasoningvision | $0.11 | DeepSeek · via Morph | 0.04 | 0.32 | 1M | - | |
| Llama 4 Scoutopen weightsvisiontools | $0.12 | Meta · via DeepInfra | 0.08 | 0.24 | 1M | - | |
| Qwen3 Coder 30B A3B Instructopen weightscodingtools | $0.12 | Alibaba · via SiliconFlow | 0.07 | 0.28 | 262K | - | |
| Qwen3 32Bopen weightsreasoningtools | $0.13 | Alibaba · via DeepInfra | 0.08 | 0.28 | 131K | - | |
| gpt-oss-safeguard-20breasoningprompt cachetools | $0.13 | OpenAI | 0.07 | 0.30 | 131K | - | |
| Qwen3 14Bopen weightsreasoningtools | $0.13 | Alibaba · via NextBit | 0.10 | 0.22 | 41K | - | |
| Qwen3 30B A3B Instruct 2507open weightsprompt cachetools | $0.14 | Alibaba · via DekaLLM | 0.09 | 0.30 | 262K | - | |
| GLM 4.7 Flashopen weightsreasoningtools | $0.15 | Zhipu · via Cloudflare | 0.06 | 0.40 | 200K | - | |
| Qwen3 235B A22B Instruct 2507open weightstools200k+ context | $0.15 | Alibaba · via GMICloud | 0.09 | 0.35 | 262K | - | |
| Llama 3.3 70B Instructopen weightstools | $0.16 | Meta · via DeepInfra | 0.10 | 0.32 | 131K | - | |
| Gemma 4 31Bopen weightsreasoningvision | $0.16 | Google · via CoreWeave | 0.10 | 0.34 | 262K | - | |
| GPT-4.1 Nanovisionprompt cachetools | $0.18 | OpenAI | 0.10 | 0.40 | 1M | - | |
| Claude Haiku 5.5reasoningvisionprompt cache | $0.20 | Anthropic · via Google | 0.10 | 0.50 | 1M | - | |
| Qwen3 30B A3Bopen weightsreasoningtools | $0.22 | Alibaba | 0.12 | 0.50 | 41K | - | |
| Qwen3.8 Flashopen weightsreasoningvision | $0.23 | Alibaba | 0.15 | 0.47 | 1M | - | |
| Qwen3.8 Omni Flashopen weightsreasoningvision | $0.23 | Alibaba | 0.15 | 0.47 | 1M | - | |
| DeepSeek V3.2open weightsreasoningprompt cache | $0.23 | DeepSeek · via GMICloud | 0.21 | 0.31 | 164K | - | |
| GPT-5.6 Lunareasoningvisionprompt cache | $0.23 | OpenAI · via OpenAI | 0.10 | 0.60 | 1M | - | |
| GPT-5.6 Luna Proreasoningvisionprompt cache | $0.23 | OpenAI · via OpenAI | 0.10 | 0.60 | 1M | - | |
| GPT-5.4 Nanoreasoningvisionprompt cache | $0.23 | OpenAI · via OpenAI | 0.10 | 0.63 | 400K | - | |
| Qwen3.5-35B-A3Bopen weightsreasoningvision | $0.24 | Alibaba · via Darkbloom | 0.07 | 0.75 | 262K | - | |
| Mistral Small 4open weightsreasoningvision | $0.26 | Mistral | 0.15 | 0.60 | 262K | - | |
| GPT-4o-minivisionprompt cachetools | $0.26 | OpenAI · via Azure | 0.15 | 0.60 | 128K | - | |
| Qwen3 VL 30B A3B Instructopen weightsvisiontools | $0.26 | Alibaba · via DeepInfra | 0.15 | 0.60 | 262K | - | |
| Qwen3 Next 80B A3B Instructopen weightsprompt cachetools | $0.27 | Alibaba · via DeepInfra | 0.07 | 0.88 | 262K | - | |
| Gemini 3.1 Flash Litereasoningvisionprompt cache | $0.28 | Google · via Google | 0.13 | 0.75 | 1M | - | |
| Qwen3 Coder Nextopen weightscodingprompt cache | $0.29 | Alibaba | 0.12 | 0.80 | 262K | - | |
| Sabaopen weightsprompt cachetools | $0.30 | Mistral | 0.20 | 0.60 | 33K | - | |
| Llama 4 Maverickopen weightsvisionprompt cache | $0.30 | Meta · via DigitalOcean | 0.19 | 0.65 | 524K | - | |
| DeepSeek V3.2 Expopen weightsreasoningtools | $0.31 | DeepSeek | 0.27 | 0.41 | 164K | - | |
| DeepSeek V4 Flash Vision Expopen weightsreasoningvision | $0.32 | DeepSeek · via DeepInfra | 0.22 | 0.65 | 1M | - | |
| GPT-5 Minireasoningvisionprompt cache | $0.34 | OpenAI · via OpenAI | 0.13 | 1.00 | 400K | - | |
| DeepSeek V3open weightstools | $0.37 | DeepSeek · via DeepInfra | 0.26 | 0.71 | 164K | - | |
| Qwen3 VL 235B A22B Instructopen weightsvisionprompt cache | $0.37 | Alibaba · via DeepInfra | 0.20 | 0.88 | 262K | - | |
| Qwen3 VL 8B Instructopen weightsvisionprompt cache | $0.38 | Alibaba | 0.25 | 0.75 | 262K | - | |
| Qwen-Plusopen weightsprompt cachetools | $0.39 | Alibaba | 0.26 | 0.78 | 1M | - | |
| Qwen3 Coder Flashopen weightscodingprompt cache | $0.39 | Alibaba | 0.20 | 0.97 | 1M | - | |
| Mistral Small 3.1 24Bopen weightsvisiontools | $0.40 | Mistral | 0.35 | 0.56 | 128K | - | |
| GLM 5.3open weightsreasoningprompt cache | $0.40 | Zhipu · via Wafer | 0.07 | 1.40 | 1M | - |
click a row to compare (up to 3) · cheapest provider shown - open a model for every route · blended = 3×in + out, /4 · colour = price band in view - OpenRouter API (live) + routes d'accès par modèle (215 modèles)
Reading this comparison
- 178 GA models tracked, from $0.04 to $67.50 on blended price (3 input : 1 output tokens) - a handful of high-end reasoning models pull the top of that range up; input and output prices also move independently, so check both.
- Output tokens cost about 12.0x input tokens on average across these models - the blended metric already weights for that.
- Cached input tokens cost about 77% less than fresh ones on average, across the 137 models that support prompt caching.
- The cheapest GA model on blended price right now is gpt-oss-120b, at $0.04/1M tokens - price alone doesn't mean it fits your task; check context, quality and modalities too.
- For your real traffic mix (request volume, input/output split, cache hit rate), the API cost calculator gives a tailored estimate instead of the blended metric.
"Blended" assumes a fixed 3:1 input-to-output token ratio for ranking purposes - your actual workload may skew differently.