Token Metrics explained

Where to find it

The Tokens tab on Spend shows input, output, cache-read and cache-write token counts for the selected period, so cache behaviour is visible separately from fresh input.

Key metrics

• •Input tokens - tokens sent to the model (your prompts + context)
• •Output tokens - tokens generated by the model
• •Cache-read tokens - input tokens served from cache. The discount is set per model: most Anthropic, OpenAI and Google models bill them at 0.10x the input rate, xAI's Grok models discount less, and a few Claude models discount more
• •Cache-write tokens - input tokens written to cache. Anthropic and the GPT 5.6 family charge 1.25x the input rate for the write; Google, xAI and earlier OpenAI models charge no write fee

Tokeven prices both with the multipliers for the exact model you called - see the model pages.

Cache hit rate

Your cache hit rate shows what percentage of input tokens were served from cache. Higher is better.

• •0-10% - no caching in place, significant savings available
• •10-30% - some natural caching from repeated system prompts
• •30%+ - active caching strategy paying off

Improving your metrics

1. Structure prompts for cache hits - put stable content (system prompts, instructions) at the beginning
2. Reuse conversation prefixes - repeated context read from cache is billed at the model's discounted cache-read rate
3. Trim conversation context - summarize prior turns instead of re-sending full history