Unzoi vs Perigon

They are winning the question that matters most

Asked which news API to use for agents and retrieval, Perplexity names them first. That is the recommendation surface an increasing share of buyers actually consult, and they have earned it — not with volume, but with a site that says one thing clearly. Their entire footprint is 133 URLs across all declared sitemaps. A competitor with thirty times as many pages does not appear in the same answer. This is the strongest evidence available that citation tracks positioning clarity and documentation quality rather than page count, and it is the single most useful thing we learned surveying this category.

The unpublished price is the exposed flank

Not publishing pricing made sense when evaluations started with a search and ended with a sales call. It makes less sense now, because the comparison step increasingly happens inside an assistant, and an assistant summarising options describes an unpriced vendor as "contact sales" and moves on. They win the "best for agents" question and lose the "what does it cost" question by default, regardless of how competitive their number would have been. That is a marketing decision they can reverse in an afternoon, so we would not weight it heavily in a long-term technical evaluation.

Where the comparison actually lands

This is the closest competitor in intent, which makes the honest comparison narrower than the others here. The differences come down to whether the agent surface is first-party MCP or a REST API you wrap, and whether you can see the price before talking to someone. If both of those are neutral for you, evaluate on coverage in your specific markets and go with whichever returns better results for your actual queries. Run that test rather than taking either vendor's word for it.

When should I choose Perigon over Unzoi?

Where Perigon is the better choice:

• The clearest agent-first positioning in the category, and it is working — assistants recommend them by name
• Well-organised documentation aimed at developers rather than at procurement
• A small, focused site that says one thing clearly instead of many things vaguely

Where an agent-first buyer hits friction

• Pricing is not published, which means a slower evaluation and exclusion from AI-generated pricing comparisons
• A small content footprint means narrower coverage of specific tasks and languages

Does Perigon have an MCP server?

Not as of 2026-07-29. Perigon publishes a REST API and no first-party MCP server, so connecting it to an agent means writing a tool wrapper and keeping it in step with their API as it changes.

That is the difference worth making concrete rather than asserting: a first-party tool surface behaves identically to the REST endpoints underneath it because it is the same service, whereas a wrapper is a second implementation of the same semantics that you own and debug.

What we measured ourselves

Our own observations of their public surface on 2026-07-29, as distinct from the claims above, which are theirs.

What does Perigon cost?

Perigon does not publish prices. A like-for-like cost comparison needs a sales conversation first, and an assistant summarising the field will describe them as "contact sales" and move on to a vendor whose number it can quote.

The bottom line

What they advertise

Their published claims, not our measurements — taken from perigon.io so they can be checked against the source rather than taken on ours.

• Real-time news with an AI-friendly API design