Pricing

from $20.00 / 1,000 research requests

Perplexity Ultra - Grounded JSON Extraction

Turn grounded web research into validated JSON with schema enforcement, source merging, confidence scoring, and batch processing.

Pricing

from $20.00 / 1,000 research requests

Rating

0.0

(0)

Developer

Chris

Actor stats

Bookmarked

Total users

Monthly active users

18 days ago

Last modified

Categories

Automation

Developer tools

Execution Mode

mode

Required

Choose 'api' to run a Standby server, or 'batch' to process a dataset.

Type:string

Default:api

Options:

apibatch

Perplexity API Key (BYOK)

perplexityApiKey

Optional

Your Perplexity (or OpenRouter) API Key. This is stored securely.

Type:string

Model Fallback Chain

modelChain

Optional

If the first model fails, the Actor will automatically retry with the next in the list.

Type:string[]

Default:

[
  "perplexity/sonar",
  "perplexity/sonar-reasoning"
]

Policy Preset

policyPreset

Optional

Apply a predefined configuration profile. ultra-fast minimizes latency, ultra-smart enables repair + debug, ultra-private enforces PII masking and zero data retention, ultra-json enables strict JSON mode.

Type:string

Default:custom

Options:

ultra-fastultra-smartultra-privateultra-jsonultra-batchultra-extractultra-debugcustom

Task Class

taskClass

Optional

Hint for the smart routing engine. 'auto' infers the task from the request shape.

Type:string

Default:auto

Options:

autochatextractionclassificationreasoningembedding

Target API Route

route

Optional

Which API route to target (usually chat for Perplexity models).

Type:string

Default:chat

Options:

chatresponsesembeddings

Max Cost Per Run (USD)

maxCostPerRunUsd

Optional

Hard budget limit. The Actor will stop processing if this cost is exceeded.

Type:number

Minimum:0

Default:5

Max Cost Per Request (USD)

maxCostPerRequestUsd

Optional

Maximum allowed cost for a single API request.

Type:number

Minimum:0

Timeout (ms)

timeoutMs

Optional

Per-model timeout in milliseconds.

Type:integer

Minimum:1000

Default:30000

Latency Target (ms)

latencyTargetMs

Optional

Target latency hint for the routing engine. Lower values prefer faster models.

Type:integer

Minimum:100

Max Retries

maxRetries

Optional

Maximum number of retries per request.

Type:integer

Minimum:0

Default:2

Enable Exact Response Cache

enableExactCache

Optional

If enabled, identical requests will be served from the internal cache (SHA-256 hash), saving tokens and money.

Type:boolean

Default:true

Cache TTL (seconds)

cacheTtlSeconds

Optional

How long cached responses remain valid.

Type:integer

Minimum:1

Default:86400

Cache Namespace

cacheNamespace

Optional

Isolate cache entries by namespace. Useful for multi-tenant setups.

Type:string

Default:default

Enable Prefix Cache

enablePrefixCache

Optional

Enable advanced prefix-based caching for long prompts.

Type:boolean

Default:true

User Scoped Cache

userScopedCache

Optional

Isolate cache by user ID (if provided in request headers).

Type:boolean

Default:true

Enable JSON Self-Healing

jsonStrictMode

Optional

Automatically strip markdown and repair broken JSON syntax from the AI's output.

Type:boolean

Default:true

JSON Schema

jsonSchema

Optional

Validate LLM output against this JSON Schema. Used with JSON strict mode.

Type:object

Enable Deterministic Repair

enableDeterministicRepair

Optional

Try regex-based JSON repair before falling back to model-based repair.

Type:boolean

Default:true

Max Repair Attempts

maxRepairAttempts

Optional

How many times to attempt repair before failing.

Type:integer

Minimum:0

Default:1

Enable Repair Model

enableRepairModel

Optional

Use a specialized model to fix JSON that deterministic repair cannot handle.

Type:boolean

Default:false

Repair Model Name

repairModel

Optional

Model to use for JSON self-healing if deterministic repair fails.

Type:string

Default:perplexity/sonar

Enable Debug Artifacts

enableDebugArtifacts

Optional

Store JSON parse/validation failure details for post-run inspection.

Type:boolean

Default:true

Enable PII Masking

piiMasking

Optional

Scrub emails and phone numbers before sending data to the AI provider.

Type:boolean

Default:false

Zero Data Retention Only

zdrOnly

Optional

Only route to providers that support zero data retention policies.

Type:boolean

Default:false

Redact Logs

redactLogs

Optional

Redact sensitive data from Actor logs.

Type:boolean

Default:true

Target Dataset ID (Batch Mode Only)

datasetId

Optional

The Apify Dataset ID containing the rows you want to process.

Type:string

Request Template (Batch Mode Only)

requestTemplate

Optional

The Perplexity payload template. Use {{fieldName}} to inject data from your dataset.

Type:object

Input Field Mapping

inputMapping

Optional

Advanced mapping of dataset fields to template variables.

Type:object

Batch Sample Size

sampleSize

Optional

Process only the first N rows (useful for testing).

Type:integer

Minimum:1

Default:10

Dead Letter Dataset ID

deadLetterDatasetId

Optional

Failed batch rows are pushed to this dataset for inspection.

Type:string

Completion Webhook URL

webhookUrl

Optional

URL to POST a summary to when batch processing completes.

Type:string

Grounded Q&A: Structured Answers with Citations

aitoolbreakdown/atb-grounded-qa

Answers a natural-language question using ONLY the URLs you provide. Returns structured JSON with per-claim citations and confidence. No hallucinated sources.

AI Tool Breakdown

Compliance-Grade Web Intelligence for AI Agents

ai_solutionist/compliance-web-intel

The scraper AI agents trust. Extract grounded facts with citations, entities, claims & RAG chunks. Built for LangChain, LlamaIndex, AutoGPT. Quality scoring, auto-citations, 6 task modes.

Jason Pellerin

Insurance Enforcement Scraper

fortuitous_pirate/insurance-enforcement-scraper

Fortuitous Pirate

AI Lead Enrichment & Outreach Generator

piotrv1001/ai-lead-enrichment-outreach-generator

Turn a CSV or Google Sheet of leads into ready-to-send outreach in one run. Scrapes each company's website, detects business signals (hiring, dev focus, B2B SaaS), and generates grounded icebreakers, angles, and value hooks with OpenAI — tied to your own value proposition and fit-scored per lead.

FalconScrape

Perplexity AI Search Scraper

zhorex/perplexity-ai-scraper

Search Perplexity AI programmatically. Get AI-generated answers with sources and citations. No API key needed. The only Perplexity scraper on Apify that works without your own API key. Built for AEO and brand monitoring. Fast, clean JSON output.

Sami

OpenRouter Ultra – Fix Broken JSON & Structured Output

chriskelch77/openrouter-ultra

Stop dealing with broken JSON from LLMs. OpenRouter Ultra delivers reliable structured outputs from text or datasets using automatic retries, fallback models, and built-in repair. Get predictable costs and production-ready results for batch processing at scale.

Chris

Perplexity AI Instant Response Actor

scraping_samurai/perplexity-ai-instant-response-actor

The Perplexity AI Actor allows you to interact with the Perplexity effortlessly. Simply input your queries and customize your results with system-level instructions and model selection. Enjoy customizable responses with no effort!

Scraping Samurai

Text-to-JSON Structured Extractor

moving_beacon-owner1/my-actor-68

A versatile Apify actor that converts unstructured text and HTML into clean, structured JSON. Supports four extraction modes with auto-detection, URL fetching, and batch processing.

Jamshaid Arif

Schema.org Markup Validator

scrappy_garden/schema-org-markup-validator

Validate Schema.org structured data for SEO. Parses JSON-LD, detects Microdata and RDFa, highlights schema types, and reports common issues like invalid JSON-LD, missing @type, non-schema.org @context, and missing key properties for popular schema types.

Bikram Adhikari

Validate Dataset(s) with JSON Schema

jaroslavhejlek/validate-dataset-with-json-schema

This Actor validates items in one or more datasets against a provided JSON Schema. Use it if you planning to add a dataset validation schema to your actor and you want test it.