Berget AI's System One model for gating agent commands: a LoRA adapter and a fine-tuned joint schema head on Cloudflare's Clef-Flash that answer noul, choice and score questions over a state in one forward pass. Trained on Swedish and English operations decisions.
A 396M ModernBERT-large cross-encoder that answers typed questions about a text or a JSON state: one or several options, scores, yes/no with "not stated", spans with evidence quotes, rankings and numeric bins, each with a confidence and an act-or-escalate signal.
Decides
noul, choice, score
choice, score, noul, rank, extract
Architecture
clef
solvi
Fine-tuned from
cloudflare/clef-flash
answerdotai/modernbert-large
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
97.0%
59.4%
Calibration error
—
0.210
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 bev and solvi?
bev is from Berget AI and solvi from solvi. Both have open weights you can download and run. Both answer noul, choice and score questions. Only solvi answers rank and extract. solvi is the smaller model, at 396M parameters to 9.0B.
Which is more accurate, bev or solvi?
They report on different suites — bev 97.0% on Berget held-out risk split (16,902 questions; same operations-traffic corpora as training), solvi 59.4% on Fast Decisions public dev split (100 rows x 17 domains, macro), the maker's own harness, zero-shot — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, bev or solvi?
bev: Free (open weights). solvi: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run bev or solvi locally?
Yes, both: systemone pull berget-ai/bev and systemone pull solvi-ai/solvi download the weights.
Evaluation suite
Berget held-out risk split (16,902 questions; same operations-traffic corpora as training)
Fast Decisions public dev split (100 rows x 17 domains, macro), the maker's own harness, zero-shot
