Qwen3-Omni-30B-A3B-Instruct
A multimodal foundation model in the Qwen series with a 30B-parameter MoE architecture that supports text, image, and audio input. Only its understanding backbone is used here.
Timeline (last 90 days)
clef-omni is based on the Qwen3-Omni-30B-A3B-Instruct base (using only the thinker, without speech output), with a total of 30B parameters, 3B activated, supporting text, JSON, images, audio, and video (sampled at 2fps), with a context of 64,000 tokens.
Aplomb 1 基于 Qwen3.5-4B Base 构建，并集成了 Qwen3-Omni-30B-A3B-Instruct 的音频编码器
All Facts (2)
clef-omni is based on the Qwen3-Omni-30B-A3B-Instruct base (using only the thinker, without speech output), with a total of 30B parameters, 3B activated, supporting text, JSON, images, audio, and video (sampled at 2fps), with a context of 64,000 tokens.
50%UnverifiedAplomb 1 基于 Qwen3.5-4B Base 构建，并集成了 Qwen3-Omni-30B-A3B-Instruct 的音频编码器
50%