Kimi position tracking
kimipositionAccount required
Reference order on Kimi tracks the order in which documents are drawn on through a long answer, which reads more like a dependency order than a ranking.
How it renders
Order of document references across the answer body.
Sampling
We report mean position with its spread, because a source that alternates between slot one and slot five is a different story from one that sits consistently at slot three.
Gotcha
Because a long answer revisits documents, a single document can occupy several positions. We return first position and citation count rather than pretending there is one slot.
What the API returns
A single collection call against Kimi for the position field, sampled n times and returned as one object per prompt.
{
"engine": "kimi",
"field": "position",
"prompt": "best ai visibility tracking tool",
"country": "GB",
"language": "en",
"n": 30,
"positions": [
{
"url": "https://nostimates.com/engines",
"avg_position": 1.8,
"best_position": 1,
"spread": 0.9
}
]
}Response fields
| Key | Type | Notes |
|---|---|---|
| url | string | Cited URL. |
| avg_position | float | Mean ordinal across samples in which it appeared. |
| best_position | integer | Best slot achieved in the sample set. |
| spread | float | Standard deviation of position across samples. |
Kimi position FAQ
- Can one document appear multiple times?
- Often. We return first position plus a count instead of a single rank.
- Is position a ranking on Kimi?
- Not really — it reflects narrative order, not preference.
Track document ordering across long Kimi answers
We will add position and repeat-citation counts to your Kimi collection.
Other Kimi data
Across every engine
Compare how position behaves on every engine we track.
Position across all engines →