Span-01
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.
One question type only, served on Respan's own POST /api/v1/scores rather than the /v1/systemone schema. Respan reports an overall F1 of 0.843 on its behaviour benchmark, against 0.715 for Jev 1.13.0 and 0.761 for Span-01 Lite. Span-01 costs $0.02 per million input tokens with output free; Span-01 Lite is free up to a daily cap. Also listed on OpenRouter (respan/span-01). Architecture and size are not published.
What it decides
- noul — answers a yes/no question with one calibrated probability
- classify — assigns a category from a fixed taxonomy
At a glance
| Maker | Respan |
| Released | 2026-09-24 |
| License | proprietary |
Hosted API
Served by Respan — $0.02/MTok input, $0/MTok output. Get access.
Read more
This page was opened by System One for Respan, who can claim the organisation and take it over at any time.