Parallel Search and Task APIs by Parallel Web Systems

# Fix list: Parallel Search and Task APIs

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/parallel-search-api, the October 2026 research run, assessed 1 October 2026. Grade BB, 74.1 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Parallel Search and Task APIs: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Security & auth, 45 out of 100, up to 9.6 more on the total

Why it scored 45: API key in `x-api-key`. The Search MCP takes a Bearer key or OAuth at /mcp-oauth. We found no key scopes (20). Every Search and Extract action is a read, and Task tools live in a separate MCP server, so the search server is a read-only subset (15). Excerpts and fetched pages are untrusted web text, and the docs index has no prompt-injection guidance (0). No audit log found (0). The privacy policy shows a SOC 2 badge and links trust.parallel.ai, which rendered nothing we could read. No security.txt per the 30 September check, and no bug bounty found (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 2. Reliability, 75 out of 100, up to 5 more on the total

Why it scored 75: incident.io page at status.parallel.ai with Search, Task, Chat, Monitor, FindAll and Entity Search components and history (20). Four incidents since July, all partial or degraded. Intermittent 4xx errors on 7 September, about an hour of Task API latency on 2 September, elevated 503s on 6 August and a prepaid billing problem on 8 July. Most have no duration listed (20). Published limits, 600 a minute for Search and Extract, 2,000 for Tasks, 300 for Chat (15). The errors page marks 429 as retryable with backoff, but there's no `Retry-After` and no idempotency guidance for creating Task runs, which the SDKs retry twice by default (10). No SLA found (0). Search, Extract and Task are GA (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 3. Agent ergonomics, 81 out of 100, up to 3.1 more on the total

Why it scored 81: The Search MCP has two tools and caps excerpts at about 25,000 characters per call (25). Result counts, excerpt settings and source policies for domains and paths (18). Errors table with a retry column, and MCP tools return structured error objects since 24 September (20). No idempotency key for Task creation, and the SDKs retry creates on 429 and 5xx. Searches are safe to repeat. MCP annotations not visible, since the server is closed (8). Python and TypeScript SDKs, but `mode` defaults to advanced, the $5 per 1,000 tier, when you leave it out (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 4. Payments & pricing, 75 out of 100, up to 3.1 more on the total

Why it scored 75: Parallel runs a separate gateway at parallelmpp.dev that takes x402 (USDC on Base) and MPP (Stripe or Tempo) for search and extract at $0.01 and an ultra task at $0.30. Nothing on api.parallel.ai, and the gateway sells one price per endpoint with no mode choice, so partial (25). Per-unit prices published without login (20). Free monthly requests and credit per the 30 September check, and the pricing page doesn't say whether a card is needed (10). The hosted Search MCP works with no key, and the gateway needs only a wallet (20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 5. Transparency & trust, 66 out of 100, up to 3 more on the total

Made of editorial 45, provenance 86.

Why it scored 66: The API is closed with published terms, and the SDKs are MIT (20). Privacy policy effective 11 August 2026. EU-endpoint search requests keep no request or response content, but there's no retention period for the default endpoint and nothing on training (12). No deprecation notices or policy found in the changelog since June (5). EU data residency for Search is documented. Subprocessors sit behind the trust centre and the privacy policy says to ask your Parallel contact (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: parallel.ai, registered 2017-12-16 (8 years) (11 of 15)
- security.txt: not found (0 of 10)

## 6. Maintenance & community, 87 out of 100, up to 1.1 more on the total

Why it scored 87: parallel-web 1.3.5 for Python on 2026-09-29 and a changelog entry on 24 September (30). Nine changelog entries since 19 August and seven Python SDK releases since 10 August (20). Closed service with a weekly changelog and support@parallel.ai. We didn't test whether support answers (12). Current Python and TypeScript SDKs, and ai.parallel/search-mcp in the registry per the 30 September check (15). The Python SDK runs lint and build in CI, plus a workflow that detects breaking changes (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 7. Schema & documentation, 94 out of 100, up to 1 more on the total

Why it scored 94: Public OpenAPI at docs.parallel.ai/public-openapi.json, linked from llms.txt (25). llms.txt and Markdown pages (10). The Search MCP docs say when to use `web_search` and that `web_fetch` follows it once candidates are narrowed, and warn that domain filters are hard filters that can cut quality. The MCP source isn't public, so we read the docs rather than the definitions (17). `mode` is an enum, and Task output schemas are JSON Schema (12). The errors page lists each code with whether to retry and what to do, and 422s carry a structured `detail` (15). A dated changelog with entries most weeks, and beta headers retire on dated entries (15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: whether the free tier needs a card, and its current size, since the pricing page we read doesn't mention it
- unchecked: SOC 2 report type and subprocessors, because trust.parallel.ai needs JavaScript
- Whether search calls that fail are billed

## Weaknesses

- Search defaults to advanced mode at $5 per 1,000 when `mode` is left out
- SDKs retry Task creation on 429 and 5xx with no idempotency key
- x402 and MPP only through parallelmpp.dev at a flat $0.01, not on api.parallel.ai
- No prompt-injection guidance, no security.txt and no public subprocessor list
- No retention period stated outside the EU endpoint

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Pass `mode: "fast"` on every search. Without it you pay the advanced rate
- Create Task runs with `max_retries=0` and check for an existing run before retrying
- Give an `objective` plus two or three short `search_queries`. That's what ranks the excerpts
- On 402, check available balance rather than total. Running tasks reserve credit
- Prefer natural-language source guidance to domain filters. The docs say filters can cut result quality

## What the review panel asked for

- Idempotency on Task creation
- fast as default
- a deprecation policy
- versioned MCP releases
- default mode to fast
- idempotency key on Task
- make mode required
- Task idempotency keys
- Add idempotency keys to Task creation
- Send Retry-After with 429s
- scoped keys
- default endpoint retention
- x402 on main host
- Stated free tier size
- publish tool definitions
- fast mode by default

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.