On November 16 Perplexity announced Perplexity Frontier Research on its official X account and a new site, research.perplexity.ai, with the goal of building a system that can run the research loop on its own. The official page names Andrew Gordon Wilson as research lead, says the effort is backed by more than 10,000 GPUs, lists six areas (multi-agent orchestration, pre-training, continual learning, reinforcement learning, systems and frontier evaluations) and commits to publishing papers, models, data and benchmarks. CEO Aravind Srinivas said on X that Perplexity will open-source its models and tooling. The team is hiring in San Francisco, Palo Alto, New York and London, with a December 5 priority deadline.
Key Takeaways
- ✓Goal: an "automated scientist", a multi-agent system that asks questions, searches wide and deep, forms hypotheses, runs experiments and keeps iterating to produce interpretable scientific understanding rather than only black-box predictions
- ✓Resources: the official page says the effort is backed by more than 10,000 GPUs; Srinivas said on X at least 10,000
- ✓Six areas: multi-agent orchestration, pre-training (open weights, new architectures, synthetic data), continual learning (curiosity, memory, persistence), reinforcement learning, systems (inference, training, runtimes, security) and frontier evaluations
- ✓The multi-agent orchestration goal is the best-quality output for whatever cost the user wants to spend, for example a small on-device model that asks a larger model for help only when it helps
- ✓Openness: papers, models, data and benchmarks will be published; the page already lists evals such as WANDR and DRACO and open weights such as Decider v1.2 (27B); December 5 is the priority application deadline
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background
The research loop of asking questions, gathering what is known, forming hypotheses, running experiments and updating has historically run inside a handful of institutions at the speed of a few people. With Frontier Research, announced on November 16, Perplexity wants to put that loop inside software that anyone can use.
How it works
The official page describes a multi-agent system: some agents search thousands of sources, some go deep on one thread, some propose experiments and others critique and hypothesize. The research is split into six areas: multi-agent orchestration, pre-training, continual learning, reinforcement learning, systems and frontier evaluations. Continual learning means models keep learning in deployment after release, and reinforcement learning runs against live retrieval and real products.
Numbers that matter
No new model or benchmark scores were released. The page lists evaluations the team has already published, such as WANDR, DRACO and DiscoverPhysics, and says new evaluations are needed that measure discovery rather than recall. Compute is more than 10,000 GPUs.
What it means for developers
What you can use today are the open releases listed on the page: Decider v1.2 (27B), contextual embedding models, MoE kernels, pplx-garden and WANDR code and data. Researchers can apply for roles in San Francisco, Palo Alto, New York or London by the December 5 priority deadline; the coming papers and models are worth watching.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.