Qwen3.5-4B with a merged rank-16 LoRA distilled from 17,408 teacher questions and 30,000 public training rows, read out as a temperature-scaled softmax over answer-letter logits. Also in 9B, 2B and GGUF.
Native MLX FP16 conversion of Laya's multilingual checkpoint (mmBERT) for Apple silicon, with a 1,024-token context. The base for multilingual fine-tunes in Laya Studio.
Decides
choice, score, noul, classify, route
choice, score, noul
Architecture
jevk5
laya
Fine-tuned from
qwen/qwen3.5-4b
convaiinnovations/laya-multilingual
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
78.4%
—
Calibration error
0.035
—
Valid action rate
—
—
Median latency
13.2 ms
—
p95 latency
—
—
Evaluation suite
JevBench public hard tier
Figures are from each model’s manifest; accuracy and latency are what the publishers report, on their own suites and hardware. Add a third model.
Questions
What is the difference between jevk5 and laya-multilingual-mlx?
jevk5 is from Alibi Serikbay and laya-multilingual-mlx from aac6fef. Both have open weights you can download and run. Both answer choice, score and noul questions. Only jevk5 answers classify and route. laya-multilingual-mlx is the smaller model, at 322M parameters to 4.0B.
Which is more accurate, jevk5 or laya-multilingual-mlx?
Only jevk5 publishes an accuracy figure (78.4% on JevBench public hard tier); laya-multilingual-mlx does not, so there is no comparison to make without your own test.
Which is cheaper, jevk5 or laya-multilingual-mlx?
jevk5: Free (open weights). laya-multilingual-mlx: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run jevk5 or laya-multilingual-mlx locally?
Yes, both: systemone pull alibi-serikbay/jevk5 and systemone pull aac6fef/laya-multilingual-mlx download the weights.
