Expand description
Cost tracking — token counting and billing for LLM calls Phase 4 feature: Production billing support
Structs§
- Context
Snapshot - Estimated input shape for one provider request. These counts are based on characters because providers do not expose tokenizers uniformly; they are intended for comparing Apollo configurations, not billing.
- Context
Summary - Cost
Record - Cost record for a single LLM call
- Cost
Summary - Cost
Tracker - Cost tracker (in-memory + persistent accounting hooks)
- Model
Cost - Cost per 1M tokens (input/output separate)
- Rate
Limit Status - Claude API rate limit status (from response headers)
- Token
Usage - Token usage for a call