How does DeepSeek's search engine see your site?
DeepSeek opened its own search index through a public API in October 2026, and it's the only AI search engine that shows its scores. Check the authority tier it gives your site and your competitors, where you rank for your queries, which of your pages a date filter would drop, and what its fetcher reads.
How DeepSeek sees
Keep this result
The share link opens this snapshot for 90 days. The CSV has every row from the four tabs, with the market and date on each.
If this tool saves you time, you can sponsor the work on GitHub.
How this works
Authority
DeepSeek gives every result an authority tier from 0 to 4, or -1 when it doesn't know, and its docs call 3 and above highly authoritative. In my tests every URL on a host carried the same tier, so the tool runs site: for each domain, 50 results, and reports the tier for each host it finds, subdomains included.
Rankings
Each query returns 50 results, sorted by DeepSeek's rerank score, which runs from 0 to 3 and combines relevance, authority and freshness. DeepSeek counts 1.5 and above as high quality. You get your best rank, your score against that line and the top five. Registries, app stores and marketplaces such as PyPI, npm, the Chrome Web Store, Apify, WordPress.org and GitHub often outrank original tool pages, and the tab says when one outranks yours.
Dates
DeepSeek gives most results a date, and its after: and before: filters work on it. In my tests a filter removed every result that had no date and never returned one outside its range, so the tool works out what a filter for the last 12 months would do from the dates on your ranking pages, without a second search. A page of yours with no date can vanish from any date filtered search. Give it a machine readable date, such as datePublished and dateModified in your structured data, then check that DeepSeek picks it up.
Fetcher
The Fetch API reads one page and returns it as Markdown. It doesn't run JavaScript, so anything a script adds after the page loads is missing. The tab also shows the fingerprint I saw on its requests, with a log search and a Cloudflare rule to find them.
Method and limitations
Treat every number as a snapshot. Tiers, scores and ranks change and depend on the market IP, and one check is one sample. Each result shows when it was taken and from which market.
DeepSeek's dates are midnight Beijing time, and its date filter works in Beijing calendar days, so the Dates tab shows dates that way.
Identical calls from the last 24 hours are reused, so a repeat within a day shows the same results and costs nothing. The fetcher fingerprint is what I observed in October 2026, not an official one, and DeepSeek can change it at any time.
Free checks run on this site's DeepSeek key, at most 10 calls each, with 3 checks an hour and 5 a day for each connection and a shared allowance for everyone each month. If that key's credit runs low, the free checks pause until it's topped up.
Checks on your own key take up to 20 queries, at most 25 calls, and are never served from the cache, so every call is fresh and goes on your DeepSeek bill. Your browser makes the calls and this site's server only sees the results, which it analyses and saves for the share link.
Check these next
- Gemini Grounding Gap Checker
Shows who Gemini cites for your query and which of its claims your page states, touches or misses.
- AI Crawler Access Checker
Probes your site as GPTBot, ClaudeBot and PerplexityBot and finds the blocks that keep you out of AI answers.
- Common Crawl Visibility Checker
Checks whether your site is in Common Crawl, with 12 months of capture history for any domain or page.