Guides
Understanding LLM costs
Billing mechanics the price lists don't explain — written from the pricing data behind our model catalog.
How LLM API pricing actually works
Input vs output tokens, cached tokens, reasoning tokens, batch tiers, context-length surcharges — the full mental model behind every LLM bill.
Aug 20, 2026
Prompt caching prices compared: OpenAI vs Anthropic vs Gemini vs DeepSeek
Every provider discounts cached input differently — and some charge for cache writes or storage. What caching really saves, with the multipliers side by side.
Aug 20, 2026
LLM batch API discounts, compared
Batch tiers cut token prices roughly in half across OpenAI, Anthropic, Gemini, and others. When your workload qualifies and what the fine print says.
Aug 20, 2026
Reasoning tokens: the invisible line on your LLM bill
Reasoning models bill their hidden thinking as output tokens. Why the same prompt can cost 5x more than the price list suggests, and how to see it.
Aug 20, 2026
How to track LLM costs per customer
Aggregate spend answers nothing. How to attribute every LLM call to a customer and feature — the event shape, the token math, and the traps.
Aug 20, 2026
List prices are the easy part.
Marginal tracks what you actually spend — every LLM call priced at that day's rates and sliced by customer, feature, or any field you define. Integration is one track() call.