Runs
Runs
Usage
What your AI models have cost, by model, by run and by chat.
Total cost
Estimated USD across every model call so far, at the rates under Prices used for estimates.
About cost estimates
Token counts are sourced from API response metadata for OpenAI, Anthropic, Gemini, Grok, DeepSeek, and the hosted providers (Mistral, Groq, OpenRouter and the rest). Ollama uses its eval_count field. AWS Bedrock does not expose token counts directly, so tokens are estimated from character length (chars ÷ 4) — costs for Bedrock calls may be less accurate. Edit prices above and click Save pricing to update rates used for future cost calculations.