Kimi brand mentions tracking
Kimi's long answers name brands in context rather than in lists, so mentions here come with substantially more surrounding text to evaluate.
How it renders
Plain text within long-form answer bodies, usually with several sentences of context.
Sampling
Mentions are more stable across samples than citations, because the model's underlying association with a category changes more slowly than its choice of sources.
Gotcha
What the API returns
A single collection call against Kimi for the brand mentions field, sampled n times and returned as one object per prompt.
{
"engine": "kimi",
"field": "brand_mentions",
"prompt": "best ai visibility tracking tool",
"country": "GB",
"language": "en",
"n": 30,
"mentions": [
{
"brand": "Nostimates",
"alias_matched": "Nostimates",
"count": 2,
"mention_rate": 0.57,
"linked": true
},
{
"brand": "Competitor",
"alias_matched": "Competitor AI",
"count": 1,
"mention_rate": 0.3,
"linked": false
}
]
}Response fields
| Key | Type | Notes |
|---|---|---|
| brand | string | Canonical brand from your configured list. |
| alias_matched | string | The exact surface form found in the answer text. |
| count | integer | Occurrences within a single answer. |
| mention_rate | float | Share of the n samples in which the brand appeared. |
| linked | boolean | Whether the mention was also a citation. |
| context | string | Surrounding sentence, for evidence export. |
What it costs
Included in the base collection. Alias lists are configured per project at no extra cost.
See credit pricing →Kimi brand mentions FAQ
- Do you normalise for answer length?
- Yes. Raw counts on Kimi are misleading without it.
- How much context do you return?
- The full surrounding passage, which on Kimi is usually enough to judge framing.
Measure Kimi mentions with length normalised
Send your brand list and we will return normalised mention share with passage context.
Other Kimi data
Across every engine
Compare how brand mentions behaves on every engine we track.
Brand mentions across all engines →