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.
Together AI's Jev-style classifier. A LoRA fine-tune of Qwen3.5-4B that reads a state, a question and 2 to 24 options and returns one option letter. Served on Together's platform; recipe published.
Decides
choice, score, noul
choice, classify, route
Architecture
djev
tev
Fine-tuned from
google/diffusiongemma-26b-a4b-it
qwen/qwen3.5-4b
License
apache-2.0
Unspecified — weights licence being finalised
Availability
Open weights + hosted API
Open weights + hosted API
Hosted by
Maisa
Together AI
Input price
$0.035/MTok
$0.042/MTok
Decision accuracy
—
88.0%
Calibration error
—
—
Valid action rate
—
—
Median latency
—
—
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 djev and tev1?
djev is from Maisa and tev1 from Together AI. Both have open weights and a hosted API. Both answer choice questions. Only djev answers score and noul. Only tev1 answers classify and route. tev1 is the smaller model, at 4.0B parameters to 26B. djev is licensed apache-2.0; tev1, other.
Which is more accurate, djev or tev1?
Only tev1 publishes an accuracy figure (88.0% on Together development set (reused, not held out)); djev does not, so there is no comparison to make without your own test.
Which is cheaper, djev or tev1?
djev: $0.035 / $0 per 1M, or free to self-host. tev1: $0.042 / $0 per 1M, or free to self-host. Open weights cost nothing per call beyond your own hardware.
Can I run djev or tev1 locally?
Yes, both: systemone pull maisa/djev and systemone pull together-ai/tev1 download the weights.
