Wouze · Open weights, local

laya-ara

laya-ara by Wouze scores 66.6 on the Arabic datasets, the only ones it ran. Calibration error 17.3%.

System One Index

–

Ran only some datasets

English

–

Arabic

66.6

Vision Index

–

Not run on images

Calibration error

17.3%

Lower is better

Median latency

54 ms

Local, not comparable

Cost per 1,000 calls

Local

As billed

Text datasets

Dataset	Accuracy	Rank	Best on the board	Calibration error
XNLI Arabic Arabic	75.3	7th of 26	89.3	18.6%
MASSIVE scenario Arabic	69.5	9th of 26	89.5	21.8%
MASSIVE intent Arabic	77.5	13th of 26	91.5	16.1%
OSACT4 offensive Arabic	88.5	10th of 26	95.5	9.1%
Tweet sentiment Arabic	49.0	22nd of 26	80.0	25.4%
MIRACL rerank Arabic	32.0	24th of 26	90.5	26.4%
MIRACL pairwise Arabic	74.5	23rd of 26	91.0	9.2%

Ranks count the models charted on the leaderboard. Every model gets the same cases and the same request, under the benchmark rules.