Overview
Text generation on the PlayKit platform — capabilities and how it works across SDKs
Text Generation
Text generation powers conversation and completion — NPC dialogue, quest text, classification, and any place your game turns a prompt into language. You send messages and get back a reply, optionally streamed token by token.
Use the default-chat-model alias for the recommended model, or default-chat-fast for a speed-optimized one — or pass a specific model name.
Capabilities
Support varies by model; each is gated by the model's capability flags.
- Reasoning effort — let capable models think harder before answering. → Reasoning
- Tools & actions — give the model functions it can call to fetch data or take actions. → Tools & Actions
- Structured output — constrain the reply to JSON matching a schema. → Structured Output
- Multimodal input — send images alongside text. → Multimodal Input
Streaming
Replies can be streamed — delivered incrementally as the model generates them — so you can show text as it arrives instead of waiting for the full response. Streaming composes with the capabilities above (for example, tool calls surface mid-stream).
In your SDK
For installation and method names, see: