A training-free layer that turns an open LLM into a decision model. It reads typed choice, yes/no and score answers from one prefill, removes option-order bias with no labels and, from a few hundred labels, calibrates or fits a closed-form head. Weights stay untouched.
A typed-decision layer for Google's DiffusionGemma, from David Villalón at Maisa AI. It compiles a request into a small answer canvas, runs one denoising read on patched vLLM and reads the probabilities of the allowed labels, for text, images and images offered as options.
Decides
choice, score, noul, classify, route
choice, score, noul
Architecture
anyjev
djev
Fine-tuned from
qwen/qwen3-8b
google/diffusiongemma-26b-a4b-it
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights + hosted API
Hosted by
—
Maisa
Input price
—
$0.035/MTok
Decision accuracy
77.1%
—
Calibration error
0.034
—
Valid action rate
—
—
Median latency
—
—
p95 latency
—
—
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 anyjev and djev?
anyjev is from Nokia and djev from Maisa. anyjev has open weights you can download and run; djev has open weights and a hosted API. Both answer choice, score and noul questions. Only anyjev answers classify and route. anyjev is the smaller model, at 8.0B parameters to 26B.
Which is more accurate, anyjev or djev?
Only anyjev publishes an accuracy figure (77.1% on LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question); djev does not, so there is no comparison to make without your own test.
Which is cheaper, anyjev or djev?
anyjev: Free (open weights). djev: $0.035 / $0 per 1M, or free to self-host. Open weights cost nothing per call beyond your own hardware.
Can I run anyjev or djev locally?
Yes, both: systemone pull nokia/anyjev and systemone pull maisa/djev download the weights.
Evaluation suite
LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question
