decision-model
Jev 1.13
Disclosure. ModelSpec is a paying TypeSafe customer. The catalogue's own pipeline sends ambiguous creator attributions to Jev and pays per input token for the answer (scripts/attribution.py, MODEL-82). No field on a TypeSafe card may be written by a Jev judgment. TypeSafe listings are attributed by code or by a person, or they are left unattributed; the judgment is refused before it is asked. Buying from a vendor earns it nothing in a ranking: see the neutrality commitment.
Type
Open weights
no
Reported benchmark scores
This card reports no benchmark scores yet, so the ranker returns it unranked, with the reason insufficient_benchmark_evidence and a null score. That is the correct answer and not a fault on the page: unranked means there is no evidence to rank on, which is a different claim from ranking last.
| Benchmark | Catalogue standing | Score |
|---|
Not applicable to this class
This card's class is decision-model. These questions have no answer for that class of model, so they are blank here and are not counted as gaps. That is a different claim from “not yet researched” below, and the published JSON keeps the two apart.
API output tok/sCapabilitiesFill-in-the-middleJSON output modeMax output tokensToken streaming
Not yet researched
A section with nothing to say does not render at all, so this is the one place the page names its gaps. Remember that null in this schema means nobody has looked, not that the model lacks the property.
6 of the 14 facts this page can show
ParametersRelease dateLast updatedTraining cutoffLineagePlatform availabilityHardware fitBenchmark scores