ZMIME

Hy4 flash Base vs Aurora2-400M-A20M

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

Model specifications
SpecificationHy4 flash BaseTencentAurora2-400M-A20MAuroraAI Research
TaskMultimodal pretrainingBase language modeling
Published capabilitiesPretrained checkpoint for continued training and adaptation. The Flash architecture uses a sparse Engram memory module.English base model pretrained on 100 billion tokens with one shared expert and 32 routed experts per layer.
Model sizeApproximately 220B total / 12B active per token400M total / approximately 19M active parameters
AccessDownloadable under the Tencent Hy LicensePublic Transformers checkpoint for completion and adaptation
SourcesModel weightsModel weights
InputsText, ImagesText
WeightsOpen weights · Tencent Hy LicenseOpen weights · Apache-2.0
ReleasedNov 18, 2026Nov 18, 2026