Price Per Token

Price Per Token is a free comparison site for large language model API costs. It lists more than 300 models from providers including OpenAI, Anthropic, Google, Mistral, DeepSeek, Meta, Qwen and IBM, showing context length, input price, output price and cached input price per million tokens. Coding, MMLU and GPQA benchmark scores sit alongside the prices, sourced from Artificial Analysis and the Hugging Face Open LLM Leaderboard, and the data is refreshed daily. Developers and teams choosing a model for an application can filter by capability, such as tool use, vision, reasoning, web search, PDF input, caching, audio input and open source weights. Provider pages show the lowest price across hosts such as OpenRouter, AWS Bedrock, Azure, Google AI Studio, Groq, Together, Fireworks, DeepInfra, SambaNova and Cerebras, with historical price trends. A cost calculator estimates spend from expected token volumes, and further tools cover model comparisons, a token counter and an in browser playground for trying models. The site also offers an MCP server that gives AI agents live pricing and benchmark data, a weekly newsletter on pricing changes and new releases, and community ratings of models in popular AI coding tools. Use of the site is free. It carries sponsored placements, and no paid plan is listed.