Text Generation
Generate natural language content using AI with flexible input methods - from simple prompts to structured message sequences.
Basic Generation
When you need to generate text content from a prompt, use .generate.text(). This is useful for creating content like explanations, summaries, or reports. Learn more about basic text generation.
const {
data: { payload }
} = await unbody.generate
.text(
"Compare Large Language Models (LLMs) with Traditional Rule-Based AI Systems analyzing: \n" +
"- Architecture and technical foundations \n" +
"- Performance metrics and scalability \n" +
"- Use cases and limitations \n" +
"- Implementation and costs \n" +
"- Future trajectories \n\n" +
"Provide specific examples and business adoption implications for each aspect."
);Generation with Options
Sometimes you need more control over the generated content. Use options like temperature and token limits to fine-tune the output style and length.
const {
data: { payload }
} = await unbody.generate
.text("Create a detailed technical specification for implementing a zero-knowledge proof system in a blockchain voting application",
{
model: "gpt-4",
topP: 0.7,
maxTokens: 1000,
temperature: 0.7,
presencePenalty: 0,
frequencyPenalty: 0
});Multi-Message Generation
When you need to send role-based messages array as a prompt, use Message-Based Input. This approach helps when providing system instructions or creating multi-turn conversations with different roles.
const {
data: { payload }
} = await unbody.generate
.text(
[
{
role: "system",
content:
"You are a helpful technical writer who explains programming concepts clearly.",
},
{
role: "user",
content:
"Explain how to implement proper error handling in a Node.js REST API. Include basic examples.",
},
],
{
model: "gpt-4",
maxTokens: 600,
temperature: 0.3,
}
);Learn more about text generation in detail in our Text Generation Guide.