Compare the true cost of running any of 137+ local LLMs (across 50 GPUs) vs hosted cloud APIs (ChatGPT, Claude, Gemini, and more). Most power users break even in 3–8 months. Cloud prices as of 2026-03-01.
| GPT-4o (OpenAI) | $2.50 / 1M input tokens + $10.00 / 1M output |
| GPT-4o Mini (OpenAI) | $0.15 / 1M input tokens + $0.60 / 1M output |
| o1 (OpenAI) | $15.00 / 1M input tokens + $60.00 / 1M output |
| o3-mini (OpenAI) | $1.10 / 1M input tokens + $4.40 / 1M output |
| Claude 3.7 Sonnet (Anthropic) | $3.00 / 1M input tokens + $15.00 / 1M output |
| Claude 3.5 Haiku (Anthropic) | $0.80 / 1M input tokens + $4.00 / 1M output |
| Claude 3 Opus (Anthropic) | $15.00 / 1M input tokens + $75.00 / 1M output |
| Gemini 2.0 Flash (Google) | $0.10 / 1M input tokens + $0.40 / 1M output |
| Gemini 1.5 Pro (Google) | $1.25 / 1M input tokens + $5.00 / 1M output |
| Gemini 1.5 Flash (Google) | $0.07 / 1M input tokens + $0.30 / 1M output |
| Llama 3.1 405B (Meta (Together.ai)) | $1.79 / 1M input tokens + $1.79 / 1M output |
| Llama 3.1 8B (Meta (Together.ai)) | $0.18 / 1M input tokens + $0.18 / 1M output |
| Mistral Large 2 (Mistral AI) | $2.00 / 1M input tokens + $6.00 / 1M output |
| Mistral Small 3 (Mistral AI) | $0.10 / 1M input tokens + $0.30 / 1M output |
| DeepSeek V3 (DeepSeek) | $0.27 / 1M input tokens + $1.10 / 1M output |
| DeepSeek R1 (DeepSeek) | $0.55 / 1M input tokens + $2.19 / 1M output |
For local inference, cost is driven by electricity use and GPU hardware amortization — typically a small fraction of a cent per 1M tokens once the GPU is paid for. The full calculator lets you pick any model from the library, any GPU, your electricity rate, and your currency (USD/EUR).