PostGrad
For subscribers

Understand the three search modes

Keyword, semantic, hybrid — what each one does, which tier unlocks it, and when to use one over another.

PostGrad's search endpoint takes a mode parameter with three values: keyword, semantic, and hybrid. They're not interchangeable — each matches different kinds of queries to different results. Picking the wrong one is the single most common source of "I got garbage back" tickets.

Before you start: you have an API key and you've made at least one call (Make your first API call).

Open the docs page on search modes

The PostGrad docs page describing the three search modes, with code samples and a comparison table.
The PostGrad docs page describing the three search modes, with code samples and a comparison table.

Keyword search: exact matches, Starter and up

A keyword search example request and response, showing exact-match hits for a specific phrase.
A keyword search example request and response, showing exact-match hits for a specific phrase.

Available on every tier starting with Starter. This is the mode that's always on.

Semantic search: intent matching, Pro and up

A semantic search example showing a query that matches entries with related meaning, not the exact same words.
A semantic search example showing a query that matches entries with related meaning, not the exact same words.

Requires Pro or Scale. If you call semantic on a Starter key, you get a 403 with code TIER_INSUFFICIENT pointing at the upgrade path.

Hybrid search: both at once, Scale only

A hybrid search example showing the combined ranked result set from keyword and semantic runs.
A hybrid search example showing the combined ranked result set from keyword and semantic runs.

Scale only. Hybrid is the most expensive mode because it's running two searches — the per-call cost reflects that.

Which mode to pick, by use case

A decision chart showing which mode to pick for common use cases: agent chat, data lookup, fuzzy recall.
A decision chart showing which mode to pick for common use cases: agent chat, data lookup, fuzzy recall.

Inspect which mode your tier supports

The Billing page showing the current subscriber tier and its associated feature set, with the three search modes listed.
The Billing page showing the current subscriber tier and its associated feature set, with the three search modes listed.

What's next

Pick the right mode, build the right query:

Connect PostGrad via MCP

Add PostGrad as an MCP server in Claude, ChatGPT, or Cursor so your agent queries knowledge directly — no REST glue code required.

Handle errors gracefully

Every error code PostGrad's API can return, what it means, and how to recover without surfacing a stack trace to your end user.

On this page

What's next