orl estimate
Bound what a batch file or a token count will cost, before anything runs, part of the orl CLI for OpenRelay GPU VMs, pods, batches and billing.
orl estimateReads a batch input (a local JSONL file, or an uploaded file id) or a token count and prices it from the public price list, which is what billing charges. Prompt tokens are bounded the way the gateway bounds them (request bytes / 3, plus 2,048 per image) and output tokens by each record's max_tokens. A standard batch bills half of each line; expedited bills the online rate. A record with openrelay.tool_config has no bound: the batch worker calls the model once per tool round.
With --max-cost it is a guard: exit 8 when the bound exceeds the cap or cannot be computed.
Flags
| Flag | Type | Required | |
|---|---|---|---|
--input-file | string | no | local batch input (JSONL, one request per line) |
--input-file-id | string | no | an uploaded batch input file id (file-...) |
--model | string | no | model id, to price a token count instead of a file |
--input-tokens | int | no | prompt tokens, with --model |
--output-tokens | int | no | output tokens, with --model |
--priority | standard, expedited | no | batch tier (a file is always a batch; a token count is online unless this is set) |
--assume-output-tokens | int | no | output tokens to assume for records that set no max_tokens (otherwise their cost has no bound) |
--max-cost | float64 | no | fail (exit 8) when the bound exceeds this many dollars |
orl embeddings create
Turn text into vectors, part of the orl CLI for OpenRelay GPU VMs, pods, batches and billing. Usage, every flag and an example.
orl files content get
Download a file's raw JSONL content, part of the orl CLI for OpenRelay GPU VMs, pods, batches and billing. Usage, every flag and an example.