Models
The Models page reads the LLM side of the bill per model, over the range picked in its header. OpenAI, Anthropic and DeepSeek accounts always appear; another source's rows appear when they carry token-metered usage, such as Alibaba Cloud Model Studio or Volcengine imports. The model is the one the bill names; a bill that names none, such as Model Studio's or Ark's, groups by its product.
- The stat row shows AI spend, total tokens, the blended $/1M (cost divided by every token, per million) and the cache share — cache tokens as a share of the input side.
- Per model lists each model's cost and its change, its input, output and cache tokens, its blended $/1M and its share of model spend. Rows with tokens but no cost still count toward the token columns.
- Tokens per day charts the daily token volume.
Findings
- A model whose cache reads are under 5% of its input side — paying input price for what could be cache price.
- One model above 70% of model spend, when two or more cost anything.
- A spend rise the two periods' token counts put on unit price rather than usage.
- A model billed with no token metering, whose unit cost cannot be computed.
Model names inside Alibaba Cloud and Volcengine billing-item text are not extracted yet.