Qwen3-Omni-Instruct#

  • Context Length: 262144

  • Model Name: Qwen3-Omni-Instruct

  • Languages: en, zh

  • Abilities: chat, vision, audio, omni, tools

  • Description: Qwen3-Omni is the natively end-to-end multilingual omni-modal foundation models. It processes text, images, audio, and video, and delivers real-time streaming responses in both text and natural speech. We introduce several architectural upgrades to improve performance and efficiency.

Specifications#

Model Spec 1 (pytorch, 30 Billion)#

  • Model Format: pytorch

  • Model Size (in billions): 30

  • Quantizations: none

  • Engines: vLLM, Transformers

  • Model ID: Qwen/Qwen3-Omni-30B-A3B-Instruct

  • Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-engine ${engine} --model-name Qwen3-Omni-Instruct --size-in-billions 30 --model-format pytorch --quantization ${quantization}

Model Spec 2 (awq, 30 Billion)#

  • Model Format: awq

  • Model Size (in billions): 30

  • Quantizations: 4bit, 8bit

  • Engines: vLLM, Transformers

  • Model ID: cpatonn/Qwen3-Omni-30B-A3B-Instruct-AWQ-{quantization}

  • Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-engine ${engine} --model-name Qwen3-Omni-Instruct --size-in-billions 30 --model-format awq --quantization ${quantization}