nodus inference chat

nodus inference chat

Send one chat completion and print the answer; the request id and charge go to stderr

Synopsis

chat sends PROMPT (or standard input, with - or no argument) as one user message and prints the answer. The request is held at its maximum cost (the input plus –max-tokens of output at the model’s rates) and charged for the tokens it used; the request id, the model that answered, the tokens and the charge are printed to stderr.

nodus inference chat [PROMPT | -] [--model MODEL | --endpoint NAME] [--max-tokens N] [flags]

Examples

  nodus inference chat "What is the capital of France?"
  nodus inference chat --model nodus/gpt-oss-20b --max-tokens 64 "Say hello"
  nodus inference chat --endpoint support-bot < question.txt
  nodus inference chat -o json "Hi" | jq .usage

Options

      --endpoint string          Send through this InferenceEndpoint of the project instead of --model
  -h, --help                     help for chat
      --idempotency-key string   Retry safely: a repeat with the same key and prompt is answered once and charged once
      --max-tokens int           Most output tokens; the request's hold is sized from it (default 512)
      --model string             Catalog model, as 'nodus inference models' lists them; nodus/indra (Indra) routes each request (default "nodus/indra")
  -o, --output string            Output format: json or yaml (the whole completion)
      --system string            System message

Options inherited from parent commands

      --base-url string   Inference origin (default: derived from the API URL)
      --context string    Context from the config file to use
      --org string        Organization (selects the context for that org)
  -p, --project string    Project to work in
  -v, --verbose           Log each API request (never credentials)

SEE ALSO

• nodus inference - Call hosted models: list them, send a chat completion and read a request’s receipt