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.
A Situated World Model for fast judgment in agent loops.
Decides
choice, score, noul
choice, route, classify
Architecture
djev
trio-spark
Fine-tuned from
google/diffusiongemma-26b-a4b-it
—
License
apache-2.0
proprietary
Availability
Open weights + hosted API
Hosted API
Hosted by
Maisa
MachineFi
Input price
$0.035/MTok
$0.042/MTok
Decision accuracy
—
—
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 trio-spark-v1?
djev is from Maisa and trio-spark-v1 from MachineFi. djev has open weights and a hosted API; trio-spark-v1 is only available as a hosted API. Both answer choice questions. Only djev answers score and noul. Only trio-spark-v1 answers route and classify. djev is licensed apache-2.0; trio-spark-v1, proprietary.
Which is more accurate, djev or trio-spark-v1?
Neither publishes an accuracy figure. Test both on your own labelled examples.
Which is cheaper, djev or trio-spark-v1?
djev: $0.035 / $0 per 1M, or free to self-host. trio-spark-v1: $0.042 / $0 per 1M. Open weights cost nothing per call beyond your own hardware.
Can I run djev or trio-spark-v1 locally?
djev yes — systemone pull maisa/djev downloads its weights. The other is only served as a hosted API.
