# Ultra-long context

ERNIE-4.5-300B-A47B is a multimodal pre-trained model based on the MoE architecture, with powerful text understanding and generation capabilities, and supports visual-language joint reasoning.

Qwen3-235B-A22B is the latest generation of Tongyi's large model series, offering a range of dense and mixture-of-experts (MoE) models. It has made breakthroughs in inference, instruction following, intelligent agent capabilities, and multilingual support.

Qwen2-VL-72B-Instruct-GGUF is a quantized version of the original model, supporting multimodal tasks and can be run through GaiaNet.