Laya typed-decisions
system-one-open · by Convai Innovations · Code and weights marked open
Base model: ModernBERT-largesourcev1.5 roster addendum A4
apache-2.0, not gated
JevBench v1.5.4 score
0.000
Option A: rank #106 of 106 ranked systems.
The three option scores and ranks are published independently; the headline is Option A.
| Option | Score | Rank |
|---|---|---|
| A · headline | 0.000 | #106 of 106 |
| B | 0.000 | #106 of 106 |
| C | 0.000 | #106 of 106 |
Published axes
- intelligence
- 0.0
- calibration
- 83.3
- speed
- 63.4
- cost
- 82.5
Bands use the published 0–100 axis values; the marked reference is Jev 1.13.0 when that axis is available.
Run and cost evidence
- Run status
- complete · 1,624 decisions · 0 missing
- Cost
- $0.0038 per 1,000 decisions · estimate · ESTIMATE: documented hosted-model estimate; no exact base-model floor applies. $0.01/M input, $0/M output; same-class hosted-encoder estimate; no exact-base market floor.
- Median latency
- 3.317 seconds, adjusted · x2 + 0.15 s (assumption, not measured)
- Endpoint condition
- evaluator-owned CPU, Sandy (AMD Ryzen 5 3600), 4 of 12 threads, nice -n 5, shared host, offline (HF_HUB_OFFLINE=1)
Model and serving disclosure
convaiinnovations/laya @ 1c5edc17a7acd8701df6fc341c0d179f1c62c982, typed-decisions/; ModernBERT-large, max_len 1024, distinct head and calibration from the English root. Fresh full run on 29 Sep; in-process, offline, shared AMD Ryzen 5 3600 CPU, four threads. Measured latency includes shared-host conditions; it is not a controlled GPU comparison. Complete valid probabilities on all types; frequent noul abstention lowers Intelligence to its zero floor and thus the composite to zero. This is not a transport failure.
Values come from the public v1.5.4 aggregate. Scores and ranks may change in a later release.
Read the full leaderboard, the v1.5.4 release page, and the published method.