Perplexity's Multiagent Coordinates 16 Agents Across Frontier AI Models
Perplexity's Multiagent, launched November 17, 2026, coordinates up to 16 agents across Claude, GPT and Gemini models for large parallel tasks, available to Max subscribers.
By Olivia Hart
3 min read

What's News
- Perplexity launched Multiagent on November 17, 2026, with up to 16 supporting agents per task.
- Claude Fable 5.2 leads a default team of Claude Opus 5.5, GPT 6.1 Sol and Gemini 4 Carbon.
- Unlike Ultra, Multiagent agents exchange findings directly rather than reporting only to the lead.
- The tier extends effort controls introduced in September 2026 (Light, Standard, High, Ultra).
- Multiagent is available to all Perplexity Max subscribers on the web.
Perplexity launched Multiagent on November 17, 2026, a new effort setting for its Computer agent platform that lets a lead agent assign up to 16 supporting agents across several frontier models to analyze complex tasks, share evidence and cross-check their work before delivering a single response. The tier sits above Ultra and is available to all Perplexity Max subscribers on the web.
The structural change is how information moves. Under Ultra, supporting agents report only to the lead. Under Multiagent, agents exchange findings directly, question assumptions and flag discrepancies for further investigation. Perplexity says the design identifies conflicting assumptions before the lead combines the team's output into the requested recommendation or deliverable. One caveat follows every promise of coordination: the lead still assembles the final answer, so the quality of the synthesis rests on a single model.
Which models run the team?
Claude Fable 5.2 serves as the lead, planning the work and dividing it among agents running Claude Opus 5.5, GPT 6.1 Sol and Gemini 4 Carbon — the default supporting trio. Fable 5.2 and GPT 6 Astra can also join as supporting agents, but only when the user requests them or a task component calls for them. In plain terms, one model directs up to 16 workers drawn from four other frontier systems.
Perplexity targets the mode at assignments with distinct but connected parts. In an annual-report review, agents can examine financial statements, footnotes and management commentary separately, then compare findings to flag inconsistent figures or disclosures for an auditor. For deal data-room diligence, separate agents can review financial records, contracts and operating documents in parallel and connect details such as a customer contract that changes the revenue outlook.
What controls do users have?
Users select Multiagent through Computer's effort slider and can name the models they want in their query. During a task, they can see which agents are active and which models those agents run. The lead remains the conversational point of contact, so users redirect the work without messaging every agent separately.
Agents also coordinate edits to shared files. Perplexity gives the example of one agent updating revenue assumptions in a financial model and notifying another to use those figures in its valuation analysis. On a follow-up, the lead tells affected agents what changed and the team continues; stopping the thread stops the entire team. The lead can also take on part of the assignment itself.
Where does Multiagent fit in Computer?
Multiagent extends the effort controls Perplexity introduced in September 2026, which offered Light, Standard, High and Ultra settings. Perplexity picked the coordinating model and reasoning level for each setting; custom controls let users choose a specific model and reasoning level themselves. Lower settings used less expensive models, and credit usage depended on the work Computer performed. What Perplexity has not disclosed is what Multiagent costs in credits relative to Ultra — a material gap for subscribers weighing the top tier.
The company draws clear boundaries between its tiers. Model Council handles judgment calls that benefit from independent perspectives: several models investigate the same question separately, then Computer combines findings and shows where they agree and disagree. Ultra suits open-ended problems requiring several approaches tested in parallel. Multiagent is for large assignments with interconnected parts that can proceed simultaneously.
The annual-report example shows the intended division of responsibility: agents compare figures and disclosures across documents, while an auditor reviews the inconsistencies they flag. Perplexity frames Multiagent's coordination as preparation for professional scrutiny — with one lead still responsible for assembling the final response, its value will depend on whether cross-checked agent work holds up under that scrutiny.
Source: unite.ai
More from Olivia Hart
Show full bio
Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.
2,301 articles