◆ Gentoo Logic · Modeling Warehouse
Warehouse › AI/Search › Perplexity

AI/Search · AWS

Perplexity

Answer engine on AWS combining a real-time search/index layer with RAG over multiple LLMs; serves its own fine-tuned Sonar models plus frontier and open models. Exposes a metered Sonar API + Agent API.

Foundation proximity: Native (own / self-hosted weights)

Foundational models & how they consume them

PrimarySonar (own, fine-tuned)· self-hosted · Native (own / self-hosted weights)
SecondaryGPT-4-class· OpenAI API · Direct API
TertiaryNVIDIA Nemotron· Agent API · Direct API

Cloud & datastore

AWSvector storeElasticsearchPostgreSQL

Compliance

SOC 2GDPR

Others that build on proprietary / self-built models

Abnormal SecurityAbridgeAdobeAI21 LabsAleph AlphaAmbience

Explore

Native (own / self-hosted weights) →More on AWS →

FAQ

Does Perplexity use a foundational AI model?
Perplexity uses Sonar (own, fine-tuned), GPT-4-class, NVIDIA Nemotron (Native (own / self-hosted weights)).
What cloud does Perplexity run on?
Perplexity runs on AWS.
What database does Perplexity use?
Perplexity uses vector store, Elasticsearch, PostgreSQL.
How close does Perplexity run to the foundational model?
Perplexity is Native (own / self-hosted weights) — close to the metal, so cheaper tokens and more control over outputs.

Sources

perplexity.ai ↗

Also on the map

proprietary / self-built modelsOpenAI · GPTNVIDIA · Nemotron

← Full orbit map (260 companies) · Score your own stack →

Architecture inferred from public sources · confidence medium · verify before betting on a detail.