Sonar Deep Research

Commission a report: this endpoint runs many searches and writes a long, cited document.

POST https://api.aisa.one/apis/v1/perplexity/sonar-deep-research

Reference example

text

# Fixed reference example. Playground inputs do not update this sample.
curl --request POST \
  --url https://api.aisa.one/apis/v1/perplexity/sonar-deep-research \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "sonar-deep-research",
  "messages": [
    {
      "role": "user",
      "content": "Write a comprehensive analysis of the global semiconductor supply chain risks in 2026"
    }
  ]
}'

200

{
  "id": "<string>",
  "model": "<string>",
  "object": "chat.completion",
  "created": 123,
  "choices": [
    {
      "index": 123,
      "message": {
        "role": "<string>",
        "content": "<string>"
      },
      "finish_reason": "stop"
    }
  ],
  "citations": [
    "<string>"
  ],
  "search_results": [
    {
      "title": "<string>",
      "url": "<string>",
      "snippet": "<string>",
      "date": "<string>",
      "source": "<string>"
    }
  ],
  "usage": {
    "prompt_tokens": 123,
    "completion_tokens": 123,
    "total_tokens": 123,
    "search_context_size": "<string>"
  }
}

Commission a report: this endpoint runs many searches and writes a long, cited document. model (required, sonar-deep-research) and messages in; choices[0].message.content, citations, search_results[] and usage out, where usage also reports num_search_queries and reasoning_tokens. ⚠️ Budget for the wait: a two-sentence question measured 192 seconds and returned 86 KB after 10 upstream searches — roughly 60 times slower and 10 times larger than post_perplexity_sonar, at the same flat per-request rate. Many clients time out well before it answers, so call it only when a report is genuinely the deliverable, and never in a loop. For anything you would read in one sitting, the other three Perplexity endpoints answer in seconds.

Example

REST

curl -X POST "https://api.aisa.one/apis/v1/perplexity/sonar-deep-research" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sonar-deep-research",
    "messages": [
      {"role": "user", "content": "Write a comprehensive analysis of the global semiconductor supply chain risks in 2026"}
    ]
  }'

First time? Point any MCP client at https://mcp.aisa.one/mcp — Claude Code, Codex, Cursor, VS Code and the rest. Authorization is OAuth: the client opens a browser, you click Allow once, and there is no key to paste. Your agent then reaches every endpoint through five tools: search, list_categories and get_details are free, use and batch_use are billed per call. Setup per client is in Connect via MCP; the live tool list and what each call costs are on aisa.one/mcp.

Set this endpoint up in your agent →

Cost and request rules

Published reference values are not an account-specific quote or a settled charge. For endpoints supporting X-AISA-Cost-Mode: quote, obtain a free quote for your exact request before execution; reconcile the settled cost with response metadata and Usage Logs. USD amounts use 1,000,000 integer micro-USD per dollar.

• Default request reference estimate (USD): 0.23932
• Cost contract: response_usd
• Published reference multiplier: 1.0
• Settlement evidence source: response_body

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

model · enum<string> · required

The Sonar model to use.

Available options: sonar, sonar-pro, sonar-reasoning-pro, sonar-deep-research

messages · object[] · required

A list of messages comprising the conversation so far.

Show child attributes

max_tokens · integer

The maximum number of tokens to generate in the response.

temperature · number · default: 0.2

Sampling temperature between 0 and 2. Lower values make output more focused and deterministic.

Required range: 0 <= x <= 2

top_p · number · default: 0.9

Nucleus sampling parameter. The model considers tokens with top_p probability mass.

Required range: 0 <= x <= 1

top_k · integer · default: 0

The number of tokens to keep for top-k filtering.

Required range: 0 <= x <= 2048

stream · boolean · default: false

Whether to stream the response using server-sent events.

search_context · enum<string> · default: low

Controls how much search context to use. Affects per-request cost.

Available options: low, medium, high

frequency_penalty · number · default: 1

Penalizes new tokens based on their existing frequency in the text so far. Positive values decrease the likelihood of repeating the same line verbatim.

Required range: 0 <= x <= 2

presence_penalty · number · default: 0

Penalizes new tokens based on whether they appear in the text so far. Positive values increase the likelihood of talking about new topics.

Required range: -2 <= x <= 2

return_citations · boolean · default: true

Whether to return citations and search results in the response.

search_recency_filter · enum<string>

Filter search results by recency.

Available options: month, week, day, hour

search_domain_filter · string[]

Limit search to specific domains.

Response

200

application/json

Successful response with AI answer and citations

id · string

Unique identifier for the completion.

model · string

The model used for the completion.

object · string

Example:

"chat.completion"

created · integer

Unix timestamp of when the completion was created.

choices · object[]

Show child attributes

citations · string[]

List of source URLs referenced in the answer.

search_results · object[]

Detailed search results with titles, snippets, and URLs.

Show child attributes

usage · object

Show child attributes