Sulat.com / Writing on AIThursday, November 12, 2026
Artificial Intelligence/12 November 2026

Parallel launches a self-serve Scale plan, with higher Search and Extract limits and Hybrid ZDR

Parallel has launched Scale, a self-serve plan between Build and Enterprise. It adds 1,500 Search and Extract requests a minute and Hybrid Zero Data Retention, billed as US$1,000 a month in API credits.

Blue-hour view of a steel-and-stone ship canal lock, with a mid-size barge moving through open gates under sodium lights. On the right, a pale lock-house wall carries a painted mural of the Parallel circular mark and the wordmark parallel.

At a glance

  • Parallel has launched Scale, a self-serve plan between pay-as-you-go Build and Enterprise, at US$1,000 a month in API credits or US$10,000 a year for US$12,000 in credits.
  • Search and Extract rise from 600 to 1,500 requests a minute. The higher Search limit excludes Turbo. Decisions stays at 600 a minute and Task at 2,000.
  • Zero data retention on Scale is Hybrid ZDR. It turns on after a signed agreement: synchronous APIs retain no request content; asynchronous APIs retain it for up to eight days.
  • Monthly credits do not roll over. You can cancel anytime; there are no prorated refunds. Strict ZDR and custom retention remain Enterprise.

Parallel has launched Scale, a self-serve plan that sits between its pay-as-you-go Build plan and Enterprise. The new tier is meant for teams that want higher production rate limits and Zero Data Retention without signing an enterprise contract.

Scale costs US$1,000 a month, before tax, and that amount is added to the organisation as credits spent at Parallel’s standard API rates, rather than as a fee on top of usage. Teams can also pay US$10,000 upfront for US$12,000 in credits over 12 months, which the company describes as two months free, or a 17% discount on the monthly price. A later note from Parallel on X points to the pricing page to compare plans.

What the higher limits actually cover

On Build, Search and Extract are capped at 600 requests a minute. Scale raises those two to 1,500 a minute. The higher Search figure excludes Turbo mode requests, according to both the launch blog and the pricing page. Decisions stays at 600 requests a minute on Build and on Scale. Task stays at 2,000 a minute on both. Enterprise limits are set in the contract.

How the credits work

The $1,000 a month isn’t a platform fee. It’s added to your organization as $1,000 in credits, which you can spend on any Parallel API at standard rates.

Parallel blog

Monthly credits cover that month and do not roll over; unused credits expire at the end of the paid month. Annual credits are drawn down over the 12-month term. Once a month’s credits are gone, further use can be paid by topping up the prepaid balance with the card on file, manually or with automatic top-ups when the balance drops below a threshold you set.

Scale is monthly by default, and you can cancel anytime. The pricing FAQs say the paid month stays covered, and that there are no prorated refunds. Plan benefits, the blog says, are kept for any months in which you’ve been charged. Billing is by card, monthly or annually upfront. Support on Scale is email, up from docs on Build.

Hybrid ZDR, after you sign

Zero data retention on Scale is Hybrid ZDR, and it is not on at the moment you pay. The blog and the pricing page say it turns on once you sign an agreement in Platform and Parallel enables it on the account. After that, synchronous APIs such as Search and Extract retain no request content. Asynchronous APIs such as Task retain request content for up to eight days.

Teams that want Strict ZDR, where asynchronous APIs keep request content for up to 60 seconds rather than eight days, or a custom retention period, still need an Enterprise plan. While ZDR is on, Parallel makes live calls to a data partner only when its agreement with that partner includes ZDR terms. Index Partners without those terms return only data Parallel already hosts, and receive no request content. Selecting a connector without ZDR terms is rejected. Bring-your-own-licence servers are not restricted.

Under Parallel’s standard Customer Terms, the company may use inputs and outputs, not linked to any individual, to train and improve the models behind its services. The docs FAQ says Enterprise plans with zero data retention exclude that use, and that Search API traffic sent to the EU endpoint is also excluded. The Scale materials describe Hybrid ZDR in terms of how long request content is kept, and do not restate that training clause.

How Scale sits next to the other plans

Build remains pay-as-you-go, with no subscription fee: you pay published API rates on a card, with up to US$80 at signup and a recurring US$5 a month, plus SOC 2 Type II, which Scale also includes. Launch is a free hosted search MCP for personal use, with variable limits, and is not an account plan covered by that certification.

Enterprise includes everything in Scale, and adds SSO (SAML), custom rate limits and data retention, a custom MSA and DPA, a HIPAA BAA, invoicing or marketplace billing, an uptime SLA, a forward-deployed engineer, a dedicated Slack channel, and early access to new products.

FeatureBuildScale
PricePay as you goUS$1,000 a month in credits, or US$10,000 a year for US$12,000 in credits
Search and Extract600 requests/min1,500 requests/min (Search excludes Turbo)
Decisions600 requests/min600 requests/min
Task2,000 requests/min2,000 requests/min
Zero data retentionNot includedHybrid ZDR, after a signed agreement
SupportDocsEmail
Build and Scale as described in Parallel’s 11 November launch blog and pricing page. SOC 2 Type II is included on both.

Published starting prices are unchanged by the plan: Search from US$1 per 1,000 requests, Extract at US$1 per 1,000 URLs, Decisions at US$1 per 1,000 requests, and higher tiers for Task, Responses, Monitor and FindAll, as set out in the pricing docs.

In Platform, the path is Billing, then Change plan, then Scale. After you pay and the credits land, you sign the ZDR agreement if you want retention turned on.


Discover more from Sulat AI & Tech News

Subscribe to get the latest posts sent to your email.