Perplexity Launches Frontier Research Unit for Autonomous Systems
The new research division will focus on multi-agent orchestration, reinforcement learning, and pre-training, backed by more than 10,000 GPUs.
Key points
- The team plans to openly release algorithms, models, data, and benchmarks tested in Perplexity products
- Research aims to build systems capable of conducting autonomous research
- Hiring researchers and engineers across San Francisco, Palo Alto, New York, and London
Why it matters
Perplexity is building dedicated in-house frontier research and compute capacity to develop and train its own models and agents.
Sources
- Perplexity Launches Frontier Research Unit for Autonomous Systems (Perplexity, Nov 16, 2026, 3:37 PM)