pinecall limits
What the org may use in this environment, what it used, and whose keys its calls run on.
pinecall limits [--prod]
What the org may use in this environment, and what it used. Every limit the deployment gave the
org (GET /v1/limits) as what is used of it; a limit with no number is no limit, and 0 refuses
everything. Minutes, messages, LLM tokens, evals, simulations and the budget (the dollars its calls
cost, sandbox and production together) are this month's and start again on the first (UTC); calls at
once, agents, seats, bought numbers, hosted apps, memory facts and knowledge chunks are what the org
holds now. A refusal past a limit is a 429 naming it, and the agent's log gets credits.exhausted,
whichever limit it was.
The last line is whose keys a call runs on: the vendors the org brought keys for
(providers add), and the ones the runtime lends it. A vendor in neither refuses
the call before it starts — bring a key for it.
$ pinecall limits
sandbox
minutes 12 of 30 · starts again 1 Nov
messages 41, no limit
LLM tokens 18210, no limit
evals 40 of 100 · starts again 1 Nov
simulations 3, no limit
budget $1.20, no limit
calls at once 0 of 1
agents 1, no limit
seats 2, no limit
bought numbers 0 of 0
hosted apps 0, no limit
memory facts 14, no limit
knowledge chunks 226, no limit
keys yours: anthropic · lent by the runtime: google, cartesia, deepgram
The MCP's usage answers the same, keys included.