Perplexity Brings Portable Computer to AMD Ryzen AI Max PCs

Perplexity expanded Portable Computer, its local-first AI agent, to systems built on AMD’s Ryzen AI Max Series processors on September 24, 2026, giving users a second hardware path beyond the NVIDIA RTX GPUs the feature required at its Windows launch ten days earlier.

Details

• New hardware support: Portable Computer now runs on AMD Ryzen AI Max Series chips, including the Ryzen AI Halo developer platform, alongside its existing NVIDIA RTX/RTX PRO support
• Hardware bar: requires 24GB or more of graphics-accessible memory, matching the threshold already set for RTX systems
• Bundled stack: Portable packages the local model, agent harness and orchestrator, scheduler, sandbox, and a sensitive-content classifier into one system, so users don’t have to assemble an inference server and tooling themselves
• Model choices: available local models include Qwen 3.8 27B and Perplexity’s own post-trained PPLX 27B
• No credit cost: inference run locally on the device does not consume Perplexity Computer credits
• Cloud escalation on request: tasks that need current information or heavier reasoning can still call out to Perplexity Search or 15+ frontier models, but only after the user approves sending data off-device
• Setup: users download a model in one click from Local Inference in the Perplexity app’s settings, which also installs the inference engine and classifier, then choose which folders and apps Portable can access
• Highlighted use cases: financial reconciliation of invoices against contracts, daily operations briefings, legal contract compliance review, and engineering pull request summaries
• Availability: live now for Pro and Max subscribers running Windows 10 or 11 with roughly 20GB of free storage

What happened next

The AMD rollout follows the same on-device privacy pitch Perplexity made when Portable Computer reached Windows on NVIDIA RTX hardware on September 14, extending the local-first option to people whose machines run AMD’s Ryzen AI Max silicon instead. Perplexity said it expects additional OEM and silicon partners to add support going forward, continuing the platform-by-platform expansion that has already covered Linux, Mac, and now two competing GPU makers on Windows.