CLI reference

The tokenpricing CLI is installed with the Python package (pip install tokenpricing). Inside the repository you can run it without installing: uv run --project libraries/python tokenpricing ....

tokenpricing pricing

Show live prices per million tokens for a model.

tokenpricing pricing MODEL [--currency CODE] [--json]

Option	Default	Description
MODEL	—	Model identifier, e.g. openai/gpt-5.2, anthropic/claude-opus-4
--currency	USD	Target currency code (e.g. EUR, GBP, CNY); uses cached FX rates
--json	off	Machine-readable output

JSON output shape:

{
  "model": "anthropic/claude-opus-4",
  "currency": "USD",
  "input_per_million": 15.0,
  "output_per_million": 75.0,
  "cache_read_per_million": 1.5,
  "cache_creation_per_million": 18.75
}

The two cache fields are null when the model does not publish cache rates — that means "no published cache pricing", not free.

tokenpricing cost

Compute the total cost of a workload from token counts.

tokenpricing cost MODEL --in INPUT --out OUTPUT [--cache-read N] [--cache-write N] [--currency CODE] [--json]

Option	Default	Description
--in	required	Input (prompt) tokens
--out	required	Output (completion) tokens
--cache-read	0	Cached input tokens served from cache hits
--cache-write	0	Input tokens written into the cache
--currency	USD	Target currency code
--json	off	Machine-readable output

tokenpricing cost openai/gpt-5.2 --in 1000 --out 500 --cache-read 250 --cache-write 100 --currency EUR

Estimating cache savings

Run cost twice — once with all tokens under --in, once split across --in and --cache-read/--cache-write — and compare the totals.

Model-id suggestions

When a model or currency is not found, the CLI suggests close matches ("Did you mean …") instead of failing silently. Model identifiers follow the canonical database keys — browse them in the dashboard or the raw JSON.