Perplexity opens Computer to Pro users with 20+ models and Slack app

• Perplexity has opened Computer beyond enterprise pilots and made it available to Pro subscribers, with the official launch post saying Pro gets access to the product while Max adds monthly credits and higher spend limits through Perplexity's launch post.
• The product pitch is a cloud agent that can use “20+ advanced models,” plus prebuilt and custom skills and “hundreds of connectors,” according to the official announcement and a parallel rollout clip from Wes Roth's demo.
• Perplexity also added a Slack app, and the Slack rollout post says teams can install it from the Marketplace, use channel context in workflows, and sync work back to Computer on the web.
• Usage is now explicitly credit-metered: a new credits screen shows a 4,000-credit Pro bonus expiring on a fixed date, while Perplexity's post only formally commits to monthly credits and higher limits for Max.

Perplexity's official launch post says Computer is now available for Pro subscribers, expanding a product that had been positioned as an enterprise “autonomous digital worker” into a broader paid tier. The company describes the package as access to “20+ advanced models,” prebuilt and custom skills, and hundreds of connectors, with Max subscribers getting extra monthly credits and higher spending ceilings.

Supporting posts fill in the product direction. Earlier launch messaging described Computer for Enterprise as an “autonomous digital worker” built for “multi-model orchestration” and professional collaboration Enterprise teaser. A third-party walkthrough says the system routes subtasks across models instead of forcing users to pick one model manually, framing it as research, coding, and deployment from one prompt multi-model breakdown. That orchestration claim is consistent with Perplexity's own emphasis on many models and skills, but the detailed routing behavior is still coming from outside observers rather than a formal spec.

The new Slack integration is the most concrete workflow change in this drop. According to the Slack app rollout, teams can install Computer through the Slack App Marketplace, build workflows using channel context, and have outputs “automatically sync” back to the web app. That turns Slack from a notification surface into a front end for invoking Computer directly inside team conversations.

The Slack screenshot shows the intended enterprise pattern: a user asks Computer to investigate abusive discount codes, pull usage data, and then connect Stripe for remediation. The demo matters less for the specific revenue-leak example than for the permissions model it implies: the agent is meant to span chat context, internal data pulls, and downstream SaaS actions from a single request. Perplexity CEO commentary echoed that positioning, with reposted remarks claiming “Slack is going to be the interface for AI in the enterprise” CEO repost.

The new usage model is visible before any API or admin docs appear. A captured usage screen shows zero plan credits for Pro, an expiring 4,000-credit bonus bucket, and an upsell promising 10,000 monthly Max credits plus a 45,000-credit upgrade bonus. That suggests Computer usage is being separated from the flat-rate chat subscription mental model and exposed as metered agent spend.

There is still some ambiguity in how execution works across tiers. Outside reporting summarized Computer as a “24/7 digital worker” powered by a Mac mini execution node, with claims about local processing for privacy and continuous task execution in enterprise setups via Rohan Paul's newsletter. Practitioner posts also show a “Welcome to Max” screen promising “Computer tasks,” “expanded rate limits,” and early access Max onboarding screenshot. For engineers, the operational question is whether Pro delivers primarily cloud-run agent workflows while the more local or always-on execution patterns remain enterprise-specific; the evidence here points in that direction, but Perplexity has not yet published a detailed public architecture note in the sources provided.