OpenAIStreamChunk

Defined in: frontend/dist/types/adapters/openai.d.ts:215

OpenAI streaming response chunk structure.

Represents a single chunk in a Server-Sent Events (SSE) stream. Each chunk contains incremental content (delta) that should be appended to build the complete message. The stream ends when finish_reason is non-null.

See

• OpenAIRequest
• OpenAIResponse
• https://platform.openai.com/docs/api-reference/chat/streaming

Example

// Content chunk during streaming
const chunk: OpenAIStreamChunk = {
  id: 'chatcmpl-123',
  object: 'chat.completion.chunk',
  created: 1677652288,
  model: 'gpt-4',
  choices: [{
    index: 0,
    delta: { content: 'Hello' },
    finish_reason: null
  }]
};

// Final chunk signaling completion
const finalChunk: OpenAIStreamChunk = {
  ...chunk,
  choices: [{
    index: 0,
    delta: {},
    finish_reason: 'stop'
  }]
};

Properties

choices

choices: object[]

Defined in: frontend/dist/types/adapters/openai.d.ts:225

Array of delta choices (usually just one)

delta

delta: object

Incremental content delta - append to previous content

delta.content?

optional content: string

Incremental text content to append

delta.role?

optional role: "assistant"

Role (only in first chunk, usually ‘assistant’)

finish_reason

finish_reason: "length" | "stop" | "tool_calls" | "content_filter" | null

Non-null when stream is complete. Indicates why generation stopped

index

index: number

Choice index in the array

---

created

created: number

Defined in: frontend/dist/types/adapters/openai.d.ts:221

Unix timestamp when the chunk was created

---

id

id: string

Defined in: frontend/dist/types/adapters/openai.d.ts:217

Unique identifier for this stream

---

model

model: string

Defined in: frontend/dist/types/adapters/openai.d.ts:223

Model generating the stream

---

object

object: "chat.completion.chunk"

Defined in: frontend/dist/types/adapters/openai.d.ts:219

Object type - always ‘chat.completion.chunk’ for streaming