Telnyx's hosted decision models, in beta on Telnyx Inference. One request carries shared state and up to 64 typed questions (choice, yes/no noul, ordered score) and returns a probability per option, with no generated text, on a TypeSafe-compatible /v1/systemone route.
A 364M Hebrew decision encoder on NeoDictaBERT: give it a state, a question and 2 to 7 free-text options and it returns a probability for each option in one forward pass, about 70 ms on a laptop CPU. Hebrew only.
Decides
choice, score, noul, classify, route
choice, classify, route
Architecture
telnyx-decision
nativ
Fine-tuned from
—
dicta-il/neodictabert
License
proprietary
cc-by-4.0
Availability
Hosted API
Open weights
Hosted by
Telnyx
—
Input price
$0.035/MTok
—
Decision accuracy
—
96.2%
Calibration error
—
0.010
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 decision-flash and nativ-he-decision?
decision-flash is from Telnyx and nativ-he-decision from Yoav Pinto. decision-flash is only available as a hosted API; nativ-he-decision has open weights you can download and run. Both answer choice, classify and route questions. Only decision-flash answers score and noul. decision-flash is licensed proprietary; nativ-he-decision, cc-by-4.0.
Which is more accurate, decision-flash or nativ-he-decision?
Only nativ-he-decision publishes an accuracy figure (96.2% on nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite)); decision-flash does not, so there is no comparison to make without your own test.
Which is cheaper, decision-flash or nativ-he-decision?
decision-flash: $0.035 / $0 per 1M. nativ-he-decision: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run decision-flash or nativ-he-decision locally?
nativ-he-decision yes — systemone pull yoav-pinto/nativ-he-decision downloads its weights. The other is only served as a hosted API.
Evaluation suite
—
nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite)
