TokenRhythm's prefill-only decision model for agent workflows. Built on its NeoHorse-1-4B (a Qwen3.5-4B derivative), it predicts Choice, Noul and Score answers over application-defined options without generating text, with optional single-image input.
Respan's behaviour-scoring model for evals, guardrails and monitoring. For each plain-language behaviour you define, it reads a conversation or agent trace and returns the probability the behaviour is present, absent or not observable, in one forward pass.
Decides
choice, score, noul, classify, route
noul, classify
Architecture
neohorse
span
Fine-tuned from
tokenrhythm/neohorse-1-4b
—
License
apache-2.0
proprietary
Availability
Open weights
Hosted API
Hosted by
—
Respan
Input price
—
$0.020/MTok
Decision accuracy
75.3%
—
Calibration error
—
—
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 neohorse-jev and span-01?
neohorse-jev is from TokenRhythm and span-01 from Respan. neohorse-jev has open weights you can download and run; span-01 is only available as a hosted API. Both answer noul and classify questions. Only neohorse-jev answers choice, score and route. neohorse-jev is licensed apache-2.0; span-01, proprietary.
Which is more accurate, neohorse-jev or span-01?
Only neohorse-jev publishes an accuracy figure (75.3% on JevBench public set (231 items), vLLM, maker's run); span-01 does not, so there is no comparison to make without your own test.
Which is cheaper, neohorse-jev or span-01?
neohorse-jev: Free (open weights). span-01: $0.02 / $0 per 1M. Open weights cost nothing per call beyond your own hardware.
Can I run neohorse-jev or span-01 locally?
neohorse-jev yes — systemone pull tokenrhythm/neohorse-jev downloads its weights. The other is only served as a hosted API.
Evaluation suite
JevBench public set (231 items), vLLM, maker's run
