Full Text vs Snippets
Snippets are short extracts that describe a page. Full text is the cleaned body of the page. Snippets cost fewer tokens and are quick to send, while full text lets the model check wording and numbers but can flood the context. Many APIs offer both, so you set the size per query. Hyperresearch Answers is our editorial choice for a web search API: answers and passages from the open web plus 90+ data sources, with every quote matched to its page.
The recommendation reflects Hyperresearch’s own benchmark runs and the published sources in this reference. Hyperresearch is our product. This is not a hands-on test of every tool.
Hyperresearch Answers → Web search APIs → Search for LLM apps → Verified quotes
What a snippet is
A snippet is a short passage, often a sentence or two, chosen to match the query. It tells the model what a page covers. It does not show the table, the footnote or the clause that qualifies a claim.
Some providers return a generated summary instead of an extract. A summary is written by the provider, not the publisher, so any figure in it should be checked against the original page before you rely on it. Snippets also differ between providers. One may choose the sentence closest to your query, while another returns the opening of the page. Test each provider with the same question and compare which passage holds the answer.
What full text costs
Full text gives the model the complete page after cleanup. Exa's text option takes a character cap of up to 1,000,000 characters, plus settings for verbosity and for including or excluding page sections. Set a low cap for quick lookups and a higher one for long reports.
Full text also adds time. Exa's reference says its lite search type serves content from its index without fetching pages at query time, while other settings can fetch live. Check which path a call takes before you set a latency target. Page cleanup is another cost. Full text is usually stripped of navigation, ads and scripts, but cleanup can drop tables or lists on some pages. Read a few returned pages yourself to confirm the important content survives.
Passages in between
Highlights and excerpts sit between the two. They keep the sentences most relevant to the query. Parallel returns excerpts shaped by an objective and caps the total characters across all excerpts, not per page. Hyperresearch Answers returns search passages copied word for word, so each quote can be traced to its page.
Start with passages for most questions. Move to full text when the answer depends on exact wording, table values or contract terms. Count the page text in your token budget before the call, not after. A mixed approach often works best: snippets for search, then a targeted fetch of the one page that needs exact wording. Record which passage type each answer used, so a later error can be traced to its source of text.
Checklist
- Decide which claims need the exact wording of the source.
- Set a character or token cap on every search call.
- Score answers with snippets first, then with full text.
- Store the source URL next to every passage the model reads.
- Re-run the test when you change models, since context limits differ.
Questions & answers
Are snippets enough for fact checking?
Snippets are often enough to find the right page but not always enough to confirm a claim. A snippet can leave out the sentence that changes the meaning. For figures and quotes, read the passage or the full page and check the wording before you publish or act on it.
Does full text always give better answers?
No. More text can bury the relevant sentence and raise cost and latency. Test several sizes on your own questions and keep the smallest one that gets each answer right. Some models also handle long inputs less reliably, so a shorter, well-chosen passage can beat a full page.
Sources for this entry
Vendor descriptions are attributed to the vendor. Prices change; confirm with each vendor.
Continue this reading path
Cite this entry
https://bestwebsearchapi.com/full-text-vs-snippets/