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.
A 0.6B replica of the System One idea trained on game environments. A Qwen3-0.6B backbone with an attention-based Choice head that scores a dynamic candidate set for Maze, Snake, ViZDoom and position prediction, shipped with its full training pipeline.
Decides
choice, score, noul, classify, route
choice, noul, score
Architecture
jevk5
nanojev
Fine-tuned from
qwen/qwen3.5-4b
qwen/qwen3-0.6b
License
apache-2.0
mit
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
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 nanojev?
jevk5 is from Alibi Serikbay and nanojev from TianyuCodings. Both have open weights you can download and run. Both answer choice, score and noul questions. Only jevk5 answers classify and route. nanojev is the smaller model, at 600M parameters to 4.0B. jevk5 is licensed apache-2.0; nanojev, mit.
Which is more accurate, jevk5 or nanojev?
Only jevk5 publishes an accuracy figure (78.4% on JevBench public hard tier); nanojev does not, so there is no comparison to make without your own test.
Which is cheaper, jevk5 or nanojev?
jevk5: Free (open weights). nanojev: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run jevk5 or nanojev locally?
Yes, both: systemone pull alibi-serikbay/jevk5 and systemone pull tianyu-codings/nanojev download the weights.
