Perplexity Introduces Photon: A Rust-Based Retrieval Engine That Cuts p99 Latency From 800 ms to 65 ms
800ms to 65ms. Perplexity's new Rust engine cuts p99 latency by 92%—and ships in production today.
Why it matters
Perplexity replaced its forked open-source retrieval stack with Photon, a custom Rust-based engine now handling all production search traffic. The latency gains (p99: 800ms → 65ms; single-call: 160ms reported) are material for API consumers and warrant adoption review by teams building on search-grounded AI products.
The key facts
11 to knowp99 latency reduced from 800ms to 65ms
Engine written in Rust; replaces forked open-source engine
Powers new Fast Search mode in Perplexity Search API
Single-call latency reported at 160ms
Handles all production traffic for Perplexity Search
Date: September 30, 2026
Single-call latency: 160ms (partial number in excerpt)
Engine written in Rust, replaces previously forked open-source engine
Handles all production traffic retrieval and ranking
New Fast Search mode available in Perplexity Search API
In-house proprietary engine (not third-party)
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Perplexity has released Photon, an in-house retrieval and ranking engine written in Rust. It replaces an open-source engine Perplexity had forked for its AI-native search stack. Photon now handles retrieval and ranking for all production traffic. It also powers a new Fast Search mode in the…