Perplexity can now split your tasks between the cloud and local AI to reduce costs

Perplexity launched Hybrid Compute on Sep 2, 2026, a feature that splits tasks between cloud models and local AI to protect private data and reduce costs. Available in the Perplexity app for Apple Silicon Macs with Pro or Max subscriptions, it supports local models including Gemma 4 E4B, Qwen 3.6, and a Perplexity post-trained Qwen 3.6, while cloud options include Claude Opus 5 and GPT 5.6 Sol.

Affiliate links on Android Authority may earn us a commission. Learn more. https://www.androidauthority.com/external-links/ Perplexity can now split your tasks between the cloud and local AI to reduce costs Sep 2, 2026 — 1:31 AM ET - Perplexity has launched Hybrid Compute, which can split sensitive tasks between cloud models and local AI. - It automatically identifies private files and information and processes them locally. - Hybrid Compute is currently available in the Perplexity app for Apple Silicon Macs. Perplexity has been releasing agentic AI tools in quick succession over the last few months, and it doesn’t look like it’s slowing down anytime soon. After releasing Personal Computer in April and Portable Computer https://www.androidauthority.com/perplexity-portable-computer-local-ai-agent-3703083/ late last month, the company is now launching Hybrid Compute. The company announced Hybrid Compute in a post on X https://x.com/perplexity ai/status/2094803515264978953 , while also explaining how it works and who it’s for. Perplexity / X https://x.com/perplexity ai/status/2094803515264978953 Think of Hybrid Compute as a middle ground between Personal Computer and Portable Computer: it splits tasks between cloud-based models and local models on your machine. That way, you get the power of cloud models for advanced reasoning or complex queries, while your private files and data remain protected by being processed locally on your device. Users can start a task as usual, and Perplexity will notify them if there’s any personal data inside the files or attachments required for the task. Users can then choose to split the task and process the private files locally while running the rest of the task in the cloud, or upload everything to the cloud anyway. Hybrid Compute supports Gemma 4 E4B, Qwen 3.6, and a Perplexity post-trained version of the Qwen 3.6 model locally. On the cloud, users can take advantage of more powerful models like Claude Opus 5 https://www.androidauthority.com/anthropic-launches-claude-opus-5-3691255/ or GPT 5.6 https://www.androidauthority.com/gpt-5-6-3685927/ Sol. This can be useful not only for tasks where the privacy of locally stored information is important but also for reducing overall usage costs. With Hybrid Compute, users will not incur token costs for processing done by their local models, which should result in a lower overall cost per task. Hybrid Compute https://www.perplexity.ai/hub/products/hybrid-compute is currently only available in the Perplexity app on Apple Silicon Macs to users with a Perplexity Pro or Max subscription. Thank you for being part of our community. Read our Comment Policy https://www.androidauthority.com/android-authority-comment-policy/ before posting.