Completion

Generate complete responses from LLM models synchronously.

Basic Completion

The complete() method generates a full response before returning:

import { igniteModel, loadModels, Message } from 'multi-llm-ts'

const config = { apiKey: process.env.OPENAI_API_KEY }
const models = await loadModels('openai', config)
const model = igniteModel('openai', models.chat[0], config)

const messages = [
  new Message('system', 'You are a helpful assistant'),
  new Message('user', 'What is the capital of France?')
]

const response = await model.complete(messages)
console.log(response.content)
// Output: "The capital of France is Paris."

Response Object

The complete() method returns a LlmResponse object:

interface LlmResponse {
  type: 'text'
  content: string           // The generated text
  finishReason?: string     // Why generation stopped

  // Token usage (if available)
  usage?: {
    prompt: number          // Input tokens
    completion: number      // Output tokens
    total: number           // Total tokens
  }
}

Finish Reasons

Common finish reasons:

• 'stop' - Natural completion (most common)
• 'length' - Hit max token limit
• 'content_filter' - Blocked by content filter
• 'tool_calls' - Stopped to execute function calls

Using Options

Customize generation with completion options:

const response = await model.complete(messages, {
  temperature: 0.7,
  maxTokens: 500,
  abortSignal: controller.signal
})

See Generation Options for all available options including temperature, maxTokens, reasoning modes, and provider-specific features.

Next Steps

• Review Generation Options for advanced parameters
• Learn about Streaming for real-time output
• Build Messages for multi-turn conversations
• Add Function Calling capabilities
• Use Structured Output for JSON
• Implement Abort Operations for cancellation