ERNIE 4.5 VL 28B A3B PT

Name: ERNIE 4.5 VL 28B A3B PT
Author: Baidu

by BaiduERNIE

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released June 2025131K context≈ 98,304 words28B params

ERNIE 4.5 VL 28B A3B PT is a Chinese-developed vision-language model that processes both text and images, with a sparse mixture-of-experts architecture activating only 3B parameters per forward pass despite its 28B total size. This makes it computationally lean relative to its parameter count. Its pretraining roots mean it sits closer to a base model, requiring further fine-tuning for polished instruction-following behavior.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Multilingual

Moderate

Coding

ERNIE 4.5 VL 28B A3B PT

by BaiduERNIE

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released June 2025131K context≈ 98,304 words28B params

ERNIE 4.5 VL 28B A3B PT is a Chinese-developed vision-language model that processes both text and images, with a sparse mixture-of-experts architecture activating only 3B parameters per forward pass despite its 28B total size. This makes it computationally lean relative to its parameter count. Its pretraining roots mean it sits closer to a base model, requiring further fine-tuning for polished instruction-following behavior.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Multilingual

Moderate

Coding

Glossary

ArchitectureThe underlying structural design of a neural network that defines how data flows through layers and components.Base ModelA pretrained model that completes text patterns but hasn't been trained to follow instructions, serving as a starting point for customization through fine-tuning.Fine-TuningThe process of further training a pre-trained model on new data to adapt it for specific tasks or domains.Forward PassA single computation cycle where input data flows through the model's layers to produce an output prediction.Instruction-FollowingThe ability of a model to understand and execute specific tasks or commands given in natural language prompts.Language ModelAn AI model trained to predict and generate text by learning patterns from large amounts of written data.Parameter CountThe total number of adjustable weights in a model; more parameters generally mean more capacity to learn, but also require more computing power.ParametersThe learned numerical values in a model — more parameters generally means more capacity but higher compute cost.PretrainingThe initial training phase where a model learns general patterns from a large dataset before being adapted for specific downstream tasks.Vision-LanguageA model designed to understand and reason about both visual content (images) and natural language text together.Vision-Language ModelAn AI model that understands both images and text, allowing it to answer questions about images or describe what it sees.

Capabilities

Capabilities

Use Case Fit

Glossary