Feature Request
Better support for batching / multi‑entity queries (e.g., multiple financial tickers) in the Perplexity API, so we can efficiently run many similar research tasks without hitting rate limits so quickly.
Problem Statement
Our production system uses the Perplexity (Sonar) API to generate brief research summaries for >1000 stock tickers.
Currently, we have to send one request per ticker (>1000 calls per refresh), which hits rate limits and creates a major performance bottleneck.
Proposed Solution
-
Provide first‑class support or clear best practices for multi‑entity queries, such as:
-
A way to pass a list of entities (tickers) and receive clearly segmented per‑ticker results.
-
Or documented, recommended batch sizes and patterns (e.g., N tickers per request) with examples and structured output (e.g., JSON array of per‑ticker summaries).
-
API Impact
-
Affects: chat completions for web‑search‑enabled models (e.g., Sonar Pro / Deep Research).
-
May require: Either a new optional parameter for entity lists, or official guidance + examples on how to safely batch multiple entities in a single prompt.
Alternatives Considered
-
One call per ticker with throttling: hits rate limits and is slow at our scale.
-
Stuffing many tickers into one prompt: responses become shallow, sometimes truncated, and hard to parse per symbol.
Additional Context
-
Universe: >1000 tickers; goal is concise, up‑to‑date qualitative summaries, not raw price data.
-
We’d also appreciate any guidance on appropriate rate‑limit tiers or enterprise options for this kind of workload.