Now Perplexity is trying to get into the local AI action
- ID
- 17913
- Status
- summarized
- Published
- 26 Aug 2026, 7:39 AM
- Fetched
- 26 Aug 2026, 9:13 AM
- Provider
- The Register
- Category
- technology
- Original URL
- https://www.theregister.com/ai-and-ml/2026/08/26/now-perplexity-is-trying-to-get-into-the-local-ai-action/5292449
- Source URL
- https://www.theregister.com/headlines.atom
Summary
- Score
- 5.5
- Created
- 26 Aug 2026, 9:14 AM
- Tags
- Audience
- developersai_agent_usersai_ml_learnerssaas_founders
What happened
Perplexity launched 'Portable Computer,' a local-first AI agent that runs on an Nvidia DGX Spark workstation with cloud inference fallback, positioning it as a way to avoid per-token API fees and keep private data on-device. The announcement name-drops Nvidia heavily amid reports Nvidia may invest $30 billion in Perplexity, and highlights small efficient models like Nvidia Nemotron 3.5 Lightning (30B), Qwen 3.6 (35B), and Qwen 3.8 (27B) as now capable of complex agentic workflows locally.
Why it matters
If you're paying rising per-token API costs for agentic workloads, the specific models Perplexity cites (Nemotron 3.5 Lightning 30B, Qwen 3.6 35B, Qwen 3.8 27B) are concrete candidates to benchmark locally before committing to DGX Spark or any local hardware. The hybrid local-cloud orchestration pattern is worth evaluating for workloads with privacy or IP constraints, though treat the Nvidia hardware pairing with skepticism given the investment context.
Discussion angle
Is the hybrid local-cloud agent pattern actually viable for Malaysian builders given DGX Spark pricing and availability here, or is this mostly a Nvidia-Perplexity marketing alignment dressed up as a cost-saving play?