AI Inference · Single-prompt mode
Docs / AI Inference

Single-prompt mode

A simpler Playground mode for a one-off text continuation instead of a back-and-forth conversation.

Single-prompt mode is a simpler way to try a model in the Playground: type one prompt, get one text continuation back — no conversation history, no back-and-forth.

When to use it

Reach for single-prompt mode when you just want to see how a model continues a piece of text — finishing a sentence, drafting a quick paragraph, or testing a prompt idea — rather than holding a multi-turn conversation. For anything conversational, use Chat instead.

Playground fields

ParameterTypeDescription
ModelrequiredselectA model from the catalog that supports single-prompt mode.
PromptrequiredtextThe text to continue.
TemperaturesliderHow random the continuation is.
Max lengthnumberCaps how long the generated continuation is allowed to be.

When to adjust temperature and max length

  • Keep temperature low when checking whether a prompt reliably produces the same kind of continuation — useful for sanity-checking a prompt template before you rely on it repeatedly.
  • Raise temperature when you want several noticeably different continuations of the same prompt, to see the range of what the model can produce.
  • Raise max length for a longer draft; lower it when you only need to glance at the first few lines a prompt produces.

What you see back

The Playground shows the generated continuation directly beneath your prompt, ready to copy.

Note
New to the Playground? Start with Chat — it covers system prompts, streaming replies, and multi-turn conversations, and works for most use cases single-prompt mode also covers.