> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scrapellm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Regional availability

> How country routing works, which regions are supported per scraper, and how to verify that a country proxy was actually applied to your request.

ScrapeLLM routes each request through infrastructure in the country you target via the `country` parameter. This page explains how to verify routing worked, how language interacts with `country`, and per-scraper region support.

## How country routing works

Pass any [ISO 3166-1 alpha-2](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) country code to the `country` parameter. ScrapeLLM routes your request through infrastructure in that region so the AI provider sees it as a local user.

```bash theme={null}
# Get ChatGPT's response as seen by users in Germany
curl "https://api.scrapellm.com/scrapers/chatgpt" \
  -H "X-API-Key: YOUR_API_KEY" \
  -G \
  --data-urlencode "prompt=Best CRM software?" \
  --data-urlencode "country=DE"
```

This is essential for localised brand research - AI responses vary significantly by region.

## Language behaviour

`country` controls geo-routing. It does **not** force a response language. AI providers infer language primarily from the prompt itself.

* Write the prompt in the target language for the most consistent localised response
* Mixed-language prompts (English prompt + non-English country) often work but may produce inconsistent localisation

## Per-scraper region support

Most scrapers support a broad set of countries. Exceptions:

| Scraper            | Unsupported countries                                     |
| ------------------ | --------------------------------------------------------- |
| **ChatGPT**        | None known                                                |
| **Perplexity**     | None known                                                |
| **Grok**           | JP, TW                                                    |
| **Copilot**        | JP, TW                                                    |
| **Gemini**         | JP, TW                                                    |
| **Google AI Mode** | JP, TW                                                    |
| **Amazon Rufus**   | Only supports: US, GB, DE, ES, FR, IT, JP, AU, IN, MX, BR |

For Amazon Rufus, the `country` parameter maps directly to Amazon marketplace:

| Country code | Marketplace   |
| ------------ | ------------- |
| `US`         | amazon.com    |
| `GB`         | amazon.co.uk  |
| `DE`         | amazon.de     |
| `ES`         | amazon.es     |
| `FR`         | amazon.fr     |
| `IT`         | amazon.it     |
| `JP`         | amazon.co.jp  |
| `AU`         | amazon.com.au |
| `IN`         | amazon.in     |
| `MX`         | amazon.com.mx |
| `BR`         | amazon.com.br |

Unknown country codes for Amazon Rufus fall back to `amazon.com`.

## Verifying the proxy was applied

A common question is: "I targeted country X but the response references country Y - is the routing broken?"

The proxy almost always applied correctly. What you're seeing is **AI model behaviour**, not a routing failure:

1. **Pre-trained bias**: If a model was trained heavily on US English content, it may default to US-centric references even when served from a different region. The proxy was applied; the model just didn't fully incorporate the regional context.
2. **Prompt overrides geography**: A prompt like "best pizza in Cincinnati" returns Cincinnati results from any proxy - the geography parameter affects the *infrastructure routing*, not what the user *asked for*.
3. **Check the cited sources**: The strongest signal that routing worked is `links[]`, `sources[]`, or `citations[]` - if these contain `.de`, `.fr`, or other country-specific domains, the proxy was applied correctly. Response language being US-centric while sources are regional is a model bias, not a routing bug.

If sources, language, and all other signals point to the wrong region, contact [hello@scrapellm.com](mailto:hello@scrapellm.com) with the affected `job_id`.

## Common questions

### Why did my US-targeted request return UK prices?

This is typically model bias from training data, not a routing failure. The AI provider processed your request from US infrastructure but generated an answer that references UK content. Check the `links[]` or `sources[]` field - if URLs are `.com` (not `.co.uk`), routing was correct.

### What countries are supported for ChatGPT?

ChatGPT supports a broad range of countries including US, GB, DE, FR, AU, CA, JP, and many more. Pass any ISO 3166-1 alpha-2 code and ScrapeLLM will route through infrastructure in that region.

### Can I get responses in a specific language without translating my prompt?

Not reliably. The AI provider determines the response language based on the prompt. Write your prompt in the target language for the most consistent results.