China engine
How to track brand visibility in Kimi
Kimi's long-context window means it ingests entire documents, so long-form B2B material that other surfaces skim gets read in full. For technical and enterprise brands in China, it is the surface where depth beats page count.
What matters on this engine
Kimi rewards a single authoritative long document over a cluster of short pages, which inverts the usual content-cluster playbook.
Citation structure
Document-level references with passage-level grounding.
Access model
Requires a genuine account. We run real accounts at deliberately low rate and rotate them, rather than farming disposable identities.
Gotcha
Data types we return for Kimi
Each field is parsed, sampled and priced separately on this engine. Open one to see how it renders here, what the response looks like and where it goes wrong.
Kimi citations
The linked sources an engine attaches to its answer.
citations
Kimi brand mentions
Unlinked names in the answer body, which citations miss entirely.
brand_mentions
Kimi sources
The full reference panel, including sources the answer never links inline.
sources
Kimi answer text
The raw synthesised answer, stored verbatim as evidence.
answer_text
Kimi share of voice
Your brand against its competitors, across the whole prompt set.
share_of_voice
Kimi sentiment
How the answer characterises a brand, not just whether it names it.
sentiment
Kimi prompt volatility
How much the same prompt disagrees with itself.
prompt_volatility
Kimi position
Where in the citation set you land, not just whether you are in it.
position
Kimi locale variance
The same prompt, run in different markets, disagreeing.
locale_variance
Track Kimi from next week
See it in the dashboard, or pull it into your own tools — either way, tell us your prompt set and target markets and we'll come back with a coverage plan and a price for this engine alongside anything else you need.
Other China engines