​

Answer Endpoint Reference

​

Canonical Docs Links

• Base docs URL: https://exa.ai/docs
• Answer reference: /reference/answer
• OpenAI-compatible guide: /reference/openai-sdk

​

Contents

• Overview
• Request shape
• When to prefer /answer
• Structured output
• Streaming
• Response shape
• Critical pitfalls

​

Overview

• run the retrieval step
• generate a direct answer or concise summary
• attach citations in one API response

​

Request Shape

POST https://api.exa.ai/answer
{
  "query": "What is the latest valuation of SpaceX?",
  "text": true
}

​

Core Parameters

​

When To Prefer /answer

• you need a grounded answer with citations
• you do not need to manually inspect or rank raw results first
• the app interface is question-driven rather than retrieval-driven

• your application needs the result list itself
• you want explicit control over contents modes on each result
• you want deeper synthesized search flows via deep or deep-reasoning

​

Structured Output

from exa_py import Exa

exa = Exa(api_key="YOUR_EXA_API_KEY")
response = exa.answer(
    "What is the latest valuation of SpaceX?",
    output_schema={
        "type": "object",
        "properties": {
            "valuation": {"type": "string"},
            "date": {"type": "string"}
        },
        "required": ["valuation", "date"]
    }
)
print(response.answer)

​

Streaming

for await (const chunk of exa.streamAnswer("Explain quantum computing")) {
  if (chunk.content) {
    process.stdout.write(chunk.content);
  }
}

​

Response Shape

• answer
• citations
• requestId
• costDollars

​

Critical Pitfalls

1. Use the answer endpoint for answer-shaped output, not as a substitute for full search result inspection.
2. If you need rich per-result extraction controls, prefer the search endpoint.
3. For SDK streaming, prefer stream_answer or streamAnswer helpers over ad hoc raw flags.