{
  "schema_version": "rangoon.article.v1",
  "id": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/",
  "url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/",
  "markdown_url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/index.md",
  "json_url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/index.json",
  "title": "Perplexity opened Deep Research to free and Pro users as a report-generating search mode",
  "summary": "Perplexity introduced Deep Research on February 14, 2025, a report-generating search mode with stated daily limits for free and Pro users.",
  "language": "en",
  "author": {
    "name": "Rangoon Editorial",
    "url": "https://rangoon.ai/insights/editorial/"
  },
  "publisher": {
    "name": "Rangoon",
    "maintainer": "Hypler",
    "url": "https://rangoon.ai/"
  },
  "format": "News analysis",
  "topic": {
    "slug": "agents",
    "label": "Agent systems",
    "url": "https://rangoon.ai/insights/topics/agents/"
  },
  "tags": [
    {
      "slug": "perplexity",
      "label": "Perplexity",
      "url": "https://rangoon.ai/insights/tags/perplexity/"
    },
    {
      "slug": "deep-research",
      "label": "deep research",
      "url": "https://rangoon.ai/insights/tags/deep-research/"
    },
    {
      "slug": "ai-research",
      "label": "AI research",
      "url": "https://rangoon.ai/insights/tags/ai-research/"
    }
  ],
  "dates": {
    "published": "2026-10-24",
    "updated": "2026-10-24",
    "event": "2025-02-14"
  },
  "event_source_id": 1,
  "retrospective": true,
  "reading_minutes": 3,
  "takeaways": [
    "Perplexity announced Deep Research on February 14, 2025 as a mode for producing in-depth reports, with up to five daily queries for non-subscribers and 500 for Pro users.",
    "The launch described an iterative search-and-reasoning workflow that read hundreds of sources, ran for roughly two to four minutes and could export reports as a PDF or document.",
    "A cited report reduces collection work but cannot establish source quality, resolve conflict, prove a vendor claim or replace review of original records used in a decision."
  ],
  "source_note": "Historical reporting: Source 0 is Perplexity’s dated Deep Research launch record. Source 1 is a dated Perplexity product update that documents a later mode-selection and notification change. It provides later operating context only; it does not redefine the February launch or prove current limits.",
  "sources": [
    {
      "id": 1,
      "title": "Perplexity: Introducing Perplexity Deep Research",
      "url": "https://www.perplexity.ai/hub/blog/introducing-perplexity-deep-research",
      "published": "2025-02-14"
    },
    {
      "id": 2,
      "title": "Perplexity Changelog: Simplified input, live information, and faster mobile apps",
      "url": "https://www.perplexity.ai/changelog/april-2025-product-update",
      "published": "2025-04-03"
    }
  ],
  "image": {
    "url": "https://rangoon.ai/assets/editorial/archive-perplexity-deep-research-february-2025-diagram-v1.webp",
    "thumbnail_url": "https://rangoon.ai/assets/editorial/archive-perplexity-deep-research-february-2025-diagram-v1-thumb.webp",
    "alt": "Original search explanatory diagram, Question to reviewed research report. Labels: Question, Search plan, Source set, Synthesis, Report, Claim review. Source boundary: dated primary release reporting; original editorial analysis.",
    "type": "original-diagram",
    "caption": "Question to reviewed research report. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied.",
    "credit": "Rangoon / Hypler",
    "source_url": "https://rangoon.ai/brand/",
    "rights_url": "https://rangoon.ai/terms/#media-use",
    "rights_note": "Hypler artwork permissions do not grant rights to third-party marks or imply endorsement."
  },
  "sections": [
    {
      "id": "research-mode-event",
      "url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/#research-mode-event",
      "heading": "A search product added a longer report-making mode",
      "paragraphs": [
        "Perplexity introduced Deep Research on February 14, 2025 as a mode intended to turn a complex question into an in-depth research report. The launch placed it inside the Perplexity web experience rather than presenting it as a general-purpose model endpoint. Perplexity said the system searched and read hundreds of sources, iteratively reasoned over what it found, then assembled a report. It described a typical run as taking about two to four minutes, which framed the feature as a longer task than an ordinary answer. That is a useful product distinction: a research pass can spend more time collecting and synthesizing material, but it does not transform a generated report into evidence that every underlying source is current, authoritative or correctly interpreted."
      ],
      "bullets": [],
      "source_ids": [
        1
      ]
    },
    {
      "id": "access-and-output",
      "url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/#access-and-output",
      "heading": "The launch tied research depth to stated usage limits",
      "paragraphs": [
        "Access was part of the February event. Perplexity said Deep Research was available to everyone, while assigning different stated daily limits: up to five queries for non-subscribers and 500 for Pro users. It also said reports could be exported as PDFs or documents, or turned into a Perplexity Page. Those details define a product workflow with a deliverable and a quota, not an unconditional right to unlimited analysis. A reader evaluating a historical claim should preserve that distinction. The release does not demonstrate that a particular account received the feature, that a report completed within a particular time, or that the output was suitable for a legal, clinical, financial or operational decision without a person examining its citations and original sources."
      ],
      "bullets": [],
      "source_ids": [
        1
      ]
    },
    {
      "id": "method-boundary",
      "url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/#method-boundary",
      "heading": "Iterative retrieval changed speed, not source authority",
      "paragraphs": [
        "Perplexity described the launch method as conducting many searches, reading documents and refining its plan as it learned. That can be useful for assembling a first map of a topic because an agentic search process can branch into related questions that a single keyword query might miss. It also creates a review problem. More pages are not automatically better evidence, and an answer can inherit errors from weak, stale or context-free pages. A user still needs to inspect whether citations actually support the nearby claims, whether important primary records were omitted, and whether a result distinguishes fact from inference. The product announcement supports the existence and stated behavior of a mode; it does not independently validate the supplier’s benchmark figures or establish accuracy for every research domain."
      ],
      "bullets": [],
      "source_ids": [
        1
      ]
    },
    {
      "id": "later-control-context",
      "url": "https://rangoon.ai/insights/archive-perplexity-deep-research-february-2025/#later-control-context",
      "heading": "Later interface changes did not change the launch record",
      "paragraphs": [
        "A Perplexity update dated April 3, 2025 provides limited later context. It said Deep Research and Pro had become one click away in the interface, that Deep Research did not retain a sticky model choice, and that web runs could send an email notification when a report was ready. These details help explain how the product was subsequently surfaced, but they should not be projected backward as February settings. The durable lesson is operational rather than promotional: a completed report should be treated as a review artifact. Preserve the question, date, cited sources and any exports that matter, then check the original publisher and version for consequential claims. That work remains necessary even when a tool finds sources faster than a person can manually search."
      ],
      "bullets": [],
      "source_ids": [
        2
      ]
    }
  ],
  "project_context": null,
  "related_links": [],
  "permissions_url": "https://rangoon.ai/permissions.txt",
  "content_hash": "sha256:c1a964556badad8a95446f9cb4b1d81ee34c7c170c4ed0e999e64948c05b0acc"
}