How to Query ChatGPT, Perplexity & Claude with a Single API Call

The Problem: Six APIs, Six Integrations

If you want to know how AI engines talk about your brand, your product category, or your competitors, you need data from multiple providers. ChatGPT gives different answers than Perplexity. Gemini surfaces different brands than Grok. Google AIO pulls from its own index. Copilot has its own biases.

Building and maintaining separate integrations for each provider means dealing with different authentication schemes, response formats, rate limits, error handling patterns, and output structures. For a single prompt, you might write 200+ lines of provider-specific code before you even get to the analysis part.

The Sellm API eliminates that complexity. One POST request, one response format, structured analysis across every provider.

The Solution: One API, Six Providers

The /v1/async-analysis endpoint accepts a prompt, a list of providers, and a list of country geos. Sellm handles the fan-out: it queries each provider in parallel, stores the raw responses, runs structured analysis (brand extraction, sentiment scoring, position tracking), and returns a unified result.

The flow works in two steps:

1. Submit - POST /v1/async-analysis returns immediately with a 202 Accepted and an analysis ID.
2. Poll - GET /v1/async-analysis/{'{'}analysisId{'}'} returns running until results are ready, then returns the full structured payload.

You can also configure a webhook to get notified when results are ready instead of polling.

Quick Start: curl

Submit a prompt to all six providers, targeting the US market:

curl -X POST https://api.sellm.io/v1/async-analysis \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "best crm for european saas teams",
    "replicates": 3,
    "providers": ["chatgpt", "perplexity", "gemini", "grok", "copilot", "google_aio"],
    "locations": ["US"]
  }'

Response (202 Accepted):

{
  "data": {
    "id": "aa_01abc",
    "projectId": "proj_123",
    "status": "running",
    "creditsReserved": 18,
    "webhook": { "configured": false, "status": null },
    "createdAt": "2026-03-14T10:00:00Z"
  }
}

The creditsReserved value is replicates x providers x locations (3 x 6 x 1 = 18). Each credit costs less than 1 cent.

Poll for results:

curl https://api.sellm.io/v1/async-analysis/aa_01abc \
  -H "Authorization: Bearer YOUR_API_KEY"

Python Example

A complete script that submits an analysis, polls until completion, and prints the results:

import requests
import time

API_KEY = "YOUR_API_KEY"
BASE_URL = "https://api.sellm.io/v1"
HEADERS = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json",
}

# Step 1: Submit the analysis
resp = requests.post(f"{BASE_URL}/async-analysis", headers=HEADERS, json={
    "prompt": "best crm for european saas teams",
    "replicates": 3,
    "providers": ["chatgpt", "perplexity", "gemini", "grok", "copilot", "google_aio"],
    "locations": ["US", "DE"],
})
resp.raise_for_status()
analysis_id = resp.json()["data"]["id"]
print(f"Submitted: {analysis_id}")

# Step 2: Poll until finished
while True:
    result = requests.get(
        f"{BASE_URL}/async-analysis/{analysis_id}",
        headers=HEADERS,
    ).json()

    status = result["data"]["status"]
    if status in ("completed", "failed"):
        break
    print(f"Status: {status}, waiting...")
    time.sleep(10)

# Step 3: Parse results
data = result["data"]
summary = data.get("summary")
if summary:
    print(f"Share of Voice: {summary['sovPct']}%")
    print(f"Average Position: {summary['avgPosition']}")
    print(f"Average Sentiment: {summary['avgSentiment']}")

# Provider breakdown
for provider_id, metrics in data.get("providerBreakdown", {}).items():
    print(f"\n{provider_id}:")
    print(f"  SOV: {metrics['sovPct']}%")
    print(f"  Position: {metrics['avgPosition']}")
    print(f"  Sentiment: {metrics['avgSentiment']}")

# Individual results
for r in data.get("results", []):
    print(f"\n[{r['provider']}] Brands mentioned: {', '.join(r['brandsMentioned'])}")
    if r.get("position") is not None:
        print(f"  Your position: {r['position']}")

JavaScript / Node.js Example

The same flow using fetch:

const API_KEY = "YOUR_API_KEY";
const BASE_URL = "https://api.sellm.io/v1";

async function analyzePrompt(prompt, providers, locations) {
  // Submit
  const submitResp = await fetch(`${BASE_URL}/async-analysis`, {
    method: "POST",
    headers: {
      Authorization: `Bearer ${API_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      prompt,
      replicates: 3,
      providers,
      locations,
    }),
  });
  const { data: submitted } = await submitResp.json();
  console.log("Submitted:", submitted.id);

  // Poll
  let result;
  while (true) {
    const pollResp = await fetch(
      `${BASE_URL}/async-analysis/${submitted.id}`,
      { headers: { Authorization: `Bearer ${API_KEY}` } }
    );
    result = await pollResp.json();

    if (["completed", "failed"].includes(result.data.status)) break;
    console.log("Waiting...");
    await new Promise((r) => setTimeout(r, 10_000));
  }

  return result.data;
}

// Usage
const data = await analyzePrompt(
  "best crm for european saas teams",
  ["chatgpt", "perplexity", "gemini", "grok", "copilot", "google_aio"],
  ["US", "DE"]
);

console.log("SOV:", data.summary?.sovPct + "%");
console.log("Providers:", Object.keys(data.providerBreakdown || {}));

Understanding the Response

When the analysis completes, the response contains three layers of data:

Summary

Top-level KPIs aggregated across all providers and replicates:

• sovPct - Share of voice: how often your brand is mentioned vs. all brands (0-100)
• avgPosition - Average position of your brand in AI responses (lower is better)
• avgSentiment - Average sentiment score toward your brand (0-1 scale)
• coveragePct - Percentage of responses that mention your brand at all
• topCompetitors - Most frequently mentioned competitor brands

Provider Breakdown

The same KPIs split by provider. This is where cross-provider comparison becomes powerful. A typical finding looks like:

{
  "providerBreakdown": {
    "chatgpt": { "sovPct": 35, "avgPosition": 2.1, "avgSentiment": 0.82 },
    "perplexity": { "sovPct": 28, "avgPosition": 3.0, "avgSentiment": 0.75 },
    "gemini": { "sovPct": 40, "avgPosition": 1.8, "avgSentiment": 0.88 },
    "grok": { "sovPct": 22, "avgPosition": 4.2, "avgSentiment": 0.70 },
    "copilot": { "sovPct": 30, "avgPosition": 2.5, "avgSentiment": 0.79 },
    "google_aio": { "sovPct": 38, "avgPosition": 2.0, "avgSentiment": 0.85 }
  }
}

This tells you immediately: your brand performs strongest on Gemini and weakest on Grok for this query. You can use this data to prioritize your optimization efforts.

Results Array

Individual analysis results for each provider-location-replicate combination. Each result contains:

• provider - Which AI engine produced this result
• country - ISO country code for the geo context
• replicateIndex - Which replicate (0-based), useful for measuring response consistency
• brandsMentioned - All brands detected in the response
• position - Your brand's position in the response (null if not mentioned)
• sentiment - Sentiment score toward your brand (0-1 scale)

Provider-Specific Analysis: Comparing Results Across Engines

The real value of querying multiple providers simultaneously is comparative analysis. Each AI engine has different training data, different retrieval mechanisms, and different biases. Here is what to look for:

• Consistency vs. divergence: If your brand ranks #1 on ChatGPT but doesn't appear on Perplexity, your content strategy may be optimized for one training set but not another.
• Sentiment gaps: High visibility with low sentiment on a specific provider suggests that the provider's training data includes negative coverage of your brand.
• Geographic variation: Running the same prompt with ["US", "DE", "FR"] reveals how AI engines treat your brand in different markets.
• Replicate stability: Multiple replicates show how consistent each provider is. High variance means the provider's response is unpredictable for that query.

Use Cases

Brand Monitoring

Run your core brand queries weekly across all providers. Track share of voice trends over time. Alert when a competitor overtakes your position on any provider.

Competitive Intelligence

Submit competitor-focused prompts ("best alternatives to [competitor]") and analyze which brands AI engines recommend. Understand where you stand in the competitive landscape across all major AI platforms.

Content Optimization

Identify queries where your brand has low coverage or poor sentiment. Use the provider breakdown to understand which engines need attention, then optimize your content accordingly.

Market Research

Query category-level prompts ("best tools for X") across multiple geos to understand how AI engines perceive your market. Discover which brands dominate AI recommendations in your space.

Pricing

Async analysis is credit-based. Each credit covers one provider-location-replicate combination. The cost is less than 1 cent per credit.

For example, analyzing a prompt across 6 providers, 2 locations, with 3 replicates uses 36 credits (6 x 2 x 3). That's roughly $0.30 for a comprehensive cross-provider analysis of a single prompt across two markets.

Credits are reserved upfront when you submit the request and are consumed as results complete.

Supported Providers

Webhooks

Instead of polling, you can provide a webhook URL when submitting the analysis. Sellm will send a signed POST request (HMAC-SHA256) to your endpoint when the analysis completes or fails. Webhook delivery is retried automatically for up to 24 hours.

curl -X POST https://api.sellm.io/v1/async-analysis \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "best crm for european saas teams",
    "replicates": 3,
    "providers": ["chatgpt", "perplexity", "gemini"],
    "locations": ["US"],
    "webhook": {
      "url": "https://yourapp.com/sellm-webhook"
    }
  }'