Fastest System One model

Latency per three-question customer-service case, clean single-job timing on four CPU cores, no GPU. Jev's figure is the API round trip from the same VM. Lower is better; every open model except OpenJEV sits between 40 ms and about one second.

Leads on p50 latency (lower is better): GLiClass edge v3.0 — 41 ms on Metatext decisions

#	Model	Decisions	Beats baseline	ECE raw → held-out fitted	p50 / case	Extraction F1
1	GLiClass edge v3.0Knowledgator criteria + single-hypothesis Noul (correct payload)	0.418	2/7	0.287	41 ms	n/a
2	GLiNER 2.5 smallFastino	0.574	4/7	0.172	132 ms†	0.812
3	LAYA multilingualConvai Innovations	0.431	2/7	0.284	149 ms†	n/a
4	GLiGuard 300MFastino	0.537	3/7	0.230	155 ms†	0.000
5	GLiNER2 baseFastino	0.631	3/7	0.350	190 ms†	0.935
6	GLiNER2 multiFastino	0.464	2/7	0.354	202 ms†	1.000
7	GLiClass base v3.0Knowledgator criteria + single-hypothesis Noul (correct payload)	0.606	5/7	0.146	300 ms	n/a
8	GLiClass instruct base v1.0Knowledgator criteria + single-hypothesis Noul (correct payload)	0.559	3/7	0.390	323 ms	n/a
9	GLiNER 2.5 baseFastino	0.643	5/7	0.251	332 ms†	0.968
10	GLiNER 2.5 multiFastino	0.494	4/7	0.244	346 ms†	0.885
11	LAYA typed-decisionsConvai Innovations	0.475	3/7	0.131	375 ms†	n/a
12	LAYAConvai Innovations	0.520	3/7	0.064	378 ms†	n/a
13	JevTypeSafehosted	0.863	6/7	0.093	441 ms†	n/a
14	GLiClass modern large v3.0Knowledgator criteria + single-hypothesis Noul (correct payload)	0.602	4/7	0.234	572 ms	n/a
15	Vonwfzyx	0.509	4/7	0.234→ 0.122	597 ms	n/a
16	GLiNER2 largeFastino	0.716	6/7	0.261	627 ms†	0.968
17	GLiClass multilang ultraKnowledgator criteria + single-hypothesis Noul (correct payload)	0.628	4/7	0.185	672 ms	n/a
18	GLiClass large v3.0Knowledgator correct payload, no prompt	0.642	5/7	0.250	977 ms	n/a
19	GLiClass instruct large v1.0Knowledgator criteria + single-hypothesis Noul (correct payload)	0.545	3/7	0.308	1.0 s	n/a
20	GLiNER multitask v1.0Knowledgator class names as labels	0.378	1/7	0.223	1.8 s	n/a
21	OpenJEVopen-jev	0.427	3/7	0.210	50 s†	n/a

Ranked by p50 latency (lower is better). Metatext decisions: Three hand-written workflows (support tickets, security incidents, agent traces), seven questions, about 123 judgments per model. Small, so treat a ±0.1 difference as a tie; every question has a majority baseline the model has to beat. Every model is zero-shot on this suite. Decisions is the task-weighted mean accuracy of the best configuration per model; ECE is the mean expected calibration error over tasks, and the arrow shows the held-out ECE after one temperature per primitive fitted on the other half. Latency is p50 per three-question customer-service case on four CPU cores (Jev: API round trip); † means taken from the benchmark run rather than a clean timing pass. Updated September 2026.

Same ranking on the other suite: typed-decisions.

Frequently asked questions

How fast is Jev?

About 440 ms per case round trip from our VM; TypeSafe quotes 70–500 ms end to end and third parties measure p50 between 127 ms and 749 ms depending on the request size.

What is the fastest open model?

GLiClass edge v3.0 at about 40 ms per case on CPU, but it is also the least accurate. Among models that beat the baseline on most questions, GLiNER 2.5 base (about 160 ms clean) is the pick.

Self-hosted Jev alternatives

Decision models with open weights and a permissive license, ranked by measured accuracy with their latency on four CPU cores, for teams that cannot send their state to a hosted API.

System One scorecard

One table, every System One model: Jev, LAYA (three checkpoints), Von, OpenJEV, seven GLiNER2 and seven GLiClass checkpoints. Decision accuracy, calibration and latency on the Metatext suite, extraction F1 where the model has a span head.

Frequently asked questions

Related