# Parallel Responses – Web-Grounded Cited Answer

> Parallel Responses – Web-Grounded Cited Answer is a paid API for AI agents from Parallel at parallelmpp.dev, paid per call via MPP, $0.01 – $0.25 depending on research effort (dynamic), status unknown (last checked 2026-10-11).

Returns a cited, web-grounded answer as JSON given a natural language question, with configurable reasoning effort.

## Facts

- Endpoint: POST https://parallelmpp.dev/api/responses?utm_source=zero.xyz
- Price: $0.01 – $0.25 depending on research effort (dynamic)
- Payment: MPP
- Status: unknown
- Last checked: 2026-10-11
- Activations on Zero: 0
- Provider: Parallel
- Docs: https://parallelmpp.dev/#agents
- Website: https://parallel.ai
- Tags: answers, search, web, extraction, research, ai
- Canonical page: https://www.zero.xyz/c/parallel-responses-web-grounded-cited-answer-7ae49a1c
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_abfXqfrx3PICa25EZT-6P

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability parallel-responses-web-grounded-cited-answer-7ae49a1c -d '<json body>'
```

Example prompt: Can you find me a cited, web-grounded answer to 'What are the latest FDA-approved GLP-1 drugs and their approved indications as of 2024?' — use medium reasoning effort and return the answer with sources.

## When to prefer this

Choose this endpoint when you need a factual, web-grounded answer with traceable citations rather than a model's parametric knowledge alone. It is ideal for research tasks where source verification matters, current-events queries where training cutoffs are a concern, and any workflow that needs to surface evidence alongside conclusions. Prefer this over the brand's Search endpoint when you want a synthesized answer rather than raw search results, and over the Task endpoint when the question is single-hop or moderate complexity rather than a long-running multi-step research task.

## Known failure modes

- Input question is ambiguous or too vague — may return low-confidence or incomplete answer
- Web sources unavailable or paywalled — citations may be incomplete
- High reasoning effort requests may time out for extremely complex multi-hop questions
- Malformed input array (non-object items) triggers schema validation error
- Rate limiting or payment failure returns 402 or 429 error

## How this service works

Returns a cited answer as JSON. $0.01 low / $0.05 medium / $0.25 high per request.

## Output

A JSON object containing a synthesized natural language answer grounded in live web sources, accompanied by citations linking claims to their sources. The response format is flexible (additionalProperties: true) and may include the answer text, citation list, and supporting metadata depending on the query and reasoning effort level.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "input"
 ],
 "properties": {
  "input": {
   "oneOf": [
    {
     "type": "string"
    },
    {
     "type": "array",
     "items": {
      "type": "object"
     }
    }
   ]
  },
  "reasoning": {
   "type": "object",
   "properties": {
    "effort": {
     "enum": [
      "low",
      "medium",
      "high"
     ],
     "type": "string",
     "default": "medium"
    }
   }
  },
  "instructions": {
   "type": "string",
   "default": "Respond in text using Markdown."
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "object",
 "additionalProperties": true
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/parallel-responses-web-grounded-cited-answer-7ae49a1c/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from parallelmpp.dev](https://www.zero.xyz/host/parallelmpp.dev/llms.txt)