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.
A Gemma 4 12B fine-tune for typed decisions that also chats and reads images. Ships as GGUF for llama.cpp, holds a 64K context on a 16 GB GPU, and serves /v1/systemone next to /v1/chat/completions.
Decides
noul, classify
choice, score, noul, classify, route
Architecture
span
winnow
Fine-tuned from
—
google/gemma-4-12b-it
License
proprietary
apache-2.0
Availability
Hosted API
Open weights
Hosted by
Respan
—
Input price
$0.020/MTok
—
Decision accuracy
—
85.7%
Calibration error
—
—
Valid action rate
—
—
Median latency
—
143 ms
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 span-01 and winnow?
span-01 is from Respan and winnow from EldanRing. span-01 is only available as a hosted API; winnow has open weights you can download and run. Both answer noul and classify questions. Only winnow answers choice, score and route. span-01 is licensed proprietary; winnow, apache-2.0.
Which is more accurate, span-01 or winnow?
Only winnow publishes an accuracy figure (85.7% on JevBench public subset (231 items), Q8_0); span-01 does not, so there is no comparison to make without your own test.
Which is cheaper, span-01 or winnow?
span-01: $0.02 / $0 per 1M. winnow: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run span-01 or winnow locally?
winnow yes — systemone pull eldanring/winnow downloads its weights. The other is only served as a hosted API.
