generateStream method
Stream a completion for prompt token by token.
When options puts tools in play the turn arrives as a single token
event: the tool-execution loop has no incremental native stream.
await for (final e in RunAnywhere.llm.generateStream('Say hello')) {
if (e is GenerationToken) stdout.write(e.text);
}
Throws SDKException into the consumer when generation fails.
Implementation
Stream<GenerationEvent> generateStream(
String prompt, {
LlmOptions? options,
}) => _stream(() => _request(prompt: prompt, options: options));