ZMIME

Ayb 0.5 0.8B vs bison-1

Compare documented capabilities and access. A dash means a comparable figure is unavailable.

Model specifications
SpecificationAyb 0.5 0.8BMetric AIbison-1Base Labs
TaskArmenian language and visionDecision scoring
Published capabilitiesLightweight Armenian text processing and image understanding. Metric AI reports 53.00% on ArmBench, evaluated zero-shot.Post-trained from Gemma 4 31B IT for classification, routing and rubric scoring.
Model size0.8B backbone tier31B parameters
AccessDownloadable weights after publisher approval on Hugging FacePublic weights and Baseten Model APIs as decision-models/baseten-bison-latest
ResultTextTyped decisions and probabilities for supplied options
Operating limitsAdapted for Armenian and English, with training focused on Eastern Armenian.Up to 131,072 tokens per question on Baseten Model APIs.
Billing termsNot published$0.10 USD per million input tokens on Baseten Model APIs.
SourcesModel weights · Additional documentationModel weights · Additional documentation
InputsText, ImagesText, Images
Context windowNot published131,072 tokens
WeightsOpen weights · CC BY-NC 4.0Open weights · MIT and Apache-2.0
ReleasedNov 20, 2026Nov 20, 2026