Perplexity has placed Deep Research inside Perplexity Computer, according to a post that first appeared on MarkTechPost. In the supplied context, Deep Research is described as breaking questions into subtasks and routing those research subtasks across more than 20 frontier models. The output formats mentioned are reports, decks, and dashboards.
Based on the provided notes, those are the core points that can be stated from the context. No date, product documentation, company statement, or technical release materials were included in the source package, so broader framing should remain attributed to MarkTechPost rather than treated as independently confirmed.
This places the item in a practical Models & Research and workflow context. It also connects to broader coverage of how AI systems are being integrated into day-to-day work, including our post on Hybrid human-AI workforces raising leadership and technical questions and our coverage of Nextdoor describing Codex with GPT-5.5 use in engineering workflows.
What the supplied context supports
From the extractor notes, the following points are supported:
- Deep Research is inside Perplexity Computer.
- Deep Research breaks questions into subtasks.
- Research subtasks are routed across 20+ frontier models.
- The output formats mentioned are reports, decks, and dashboards.
- The post appeared first on MarkTechPost.
The source article title provided in the notes is: “Perplexity Moves Deep Research Into Computer, Routing Research Subtasks Across 20+ Frontier Models For Reports, Decks, And Dashboards.”
What remains source-attributed
The supplied context also says some claims cannot be independently verified from the materials provided here. Those include:
- MarkTechPost's characterization that Perplexity "moved" Deep Research into Computer.
- MarkTechPost's framing that Deep Research routes research subtasks across 20+ frontier models for reports, decks, and dashboards.
Because no official announcement or product page was included in the source package, this report does not go beyond those attributed claims.
The limited context suggests a system focused on task decomposition and multi-model routing rather than a single-model response. That makes the item relevant to readers tracking AI tooling and production workflows, including cost and orchestration questions discussed in Rising AI costs prompting tighter review of marketing workflows.
For background on multi-agent and multi-step prompting patterns, see Google's Prompting Guide and the Stanford paper on HELM: Holistic Evaluation of Language Models, which helps frame why model selection and orchestration can matter in practical use. Readers looking for official product context can also monitor Perplexity's official site.
The supplied context does not include:
- a release date
- a company statement
- a product page
- technical documentation
- a list of the 20+ models
- routing criteria
- export, editing, pricing, or availability details
As a result, the article should be read narrowly: MarkTechPost reported that Deep Research is inside Perplexity Computer, that questions are split into subtasks, that those subtasks are routed across 20+ frontier models, and that the outputs mentioned are reports, decks, and dashboards.
This report is therefore best understood as a source-attributed summary of a tools-and-workflows development, not a fully documented product analysis.
