AI token usage statistics
| Property |
Type |
Description |
cached_tokens? |
number |
Subset of prompt_tokens served from the provider’s prompt cache (OpenAI automatic prefix caching, Anthropic prompt caching). 0 when caching didn’t fire or the provider doesn’t report it. (Ask 4) |
completion_tokens |
number |
- |
prompt_tokens |
number |
- |
total_tokens? |
number |
- |
