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

# Choosing a proxy mode

> A practical decision guide: when to use standard vs premium for different sites.

Reader has two proxy modes: `standard` (the default, 1 credit per page) and `premium` (3 credits per page, uses residential proxies to bypass bot walls). Most scrapes work fine with standard. This guide is for deciding when to upgrade to premium.

## The default: `standard`

`standard` uses datacenter proxies and works for the vast majority of the web: blogs, documentation sites, news, public APIs, marketing pages, and most government or academic sites. If you have no reason to think a site is hostile, leave `proxyMode` unset and it defaults to `standard`.

```ts theme={null}
await reader.read({ url }); // implicit proxyMode: "standard"
```

## When to force `premium`

Force `proxyMode: "premium"` when:

* **The target is known-hostile.** Amazon, LinkedIn, booking sites, ticket resellers, most aggressive e-commerce. These sites block datacenter IP ranges entirely, so standard mode will always fail. Go straight to premium and get clean data on the first try.
* **Standard is returning thin or wrong content.** If `metadata.statusCode` is 200 but the markdown is very short, or contains phrases like "checking your browser" or "access denied", the site is blocking standard mode. Switch to premium.
* **You need consistent results across a batch.** Some sites block standard inconsistently: some pages work, others don't. Forcing premium gives you a uniform experience across the run.

```ts theme={null}
await reader.read({
  url: "https://www.amazon.com/dp/B08N5WRWNW",
  proxyMode: "premium",
});
```

## When to stick with `standard`

Use `standard` explicitly (or leave `proxyMode` unset) when:

* **You know the target is friendly.** Your own blog, a partner's public docs, a government site, a content API. No need to pay 3x.
* **You want a cost guarantee.** With `standard` explicit, every scrape is exactly 1 credit. For a 10,000-URL batch where budget certainty matters, lock the mode.
* **You want an error signal for blocks.** If you force `standard` and the target blocks you, Reader returns an error rather than silently costing more. You find out immediately that something changed.

## Per-site cheat sheet

| Site type                              | Recommended mode                                       |
| -------------------------------------- | ------------------------------------------------------ |
| Blogs, marketing sites, personal sites | `standard`                                             |
| Documentation sites (most)             | `standard`                                             |
| News sites (most)                      | `standard`                                             |
| Government, academic                   | `standard`                                             |
| Public APIs (JSON endpoints)           | `standard`                                             |
| General e-commerce                     | `standard` (try first; switch to `premium` if blocked) |
| Amazon product pages                   | `premium`                                              |
| LinkedIn                               | `premium`                                              |
| Booking / travel sites                 | `premium`                                              |
| Social media (where accessible)        | `premium`                                              |
| Ticketing sites (StubHub, etc.)        | `premium`                                              |

## Per-URL split in a batch

`proxyMode` applies to the whole batch, not per URL. If you have a mixed list of friendly and hostile URLs, two options:

1. **Pre-split the batch.** Route friendly URLs to a `standard` batch and hostile URLs to a `premium` batch.
2. **Run all hostile URLs at `premium`.** If you know certain domains always need premium, use a lookup to route them to a separate call.

Option 1 keeps costs predictable. For small mixed batches, running everything at `premium` is simpler if cost is less of a concern.

## Checking what mode was used

Every response includes `metadata.proxyMode` so you can see exactly what ran:

```ts theme={null}
const result = await reader.read({ url });
if (result.kind === "scrape") {
  console.log(result.data.metadata.proxyMode); // "standard" or "premium"
}
```

For a batch, check the distribution across results to understand which sites needed premium:

```ts theme={null}
const premiumCount = result.data.results.filter((r) => r.proxyMode === "premium").length;
const total = result.data.results.length;
console.log(`Premium rate: ${((premiumCount / total) * 100).toFixed(0)}%`);
```

## Next

* [Proxy modes (concept)](/home/concepts/proxy-modes)
* [Credits and billing](/home/concepts/credits-and-billing)
* [Cost estimation](/home/guides/production/cost-estimation)