Capability gains

Alibaba releases open real-time Qwen2.5-Omni model

Alibaba's Qwen team released Qwen2.5-Omni-7B, an open end-to-end model that consumes text, images, audio and video while streaming both text and natural speech.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

CURRENT ASSESSMENT · REVISION 1
TOWARD DOOM58confidence 97/100

Why it moved the index

The open checkpoint unified real-time perception and speech generation across four input modalities, materially broadening deployable multimodal agents and reducing centralized control.

AUDIT TRAIL

Assessment history

  1. R1
    Toward 58 · confidence 97

    New March 2025 open end-to-end multimodal model with no durable identity or event collision.

    12 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Alibaba releases open real-time Qwen2.5-Omni model.
  1. DoomBench assesses “Alibaba releases open real-time Qwen2.5-Omni model” as evidence moving toward doom, with magnitude 58 and confidence 97 out of 100 in the capability gains category.

  2. The DoomBench assessment of “Alibaba releases open real-time Qwen2.5-Omni model” is based on reporting from Qwen and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “Alibaba releases open real-time Qwen2.5-Omni model” as follows: Alibaba's Qwen team released Qwen2.5-Omni-7B, an open end-to-end model that consumes text, images, audio and video while streaming both text and...

    https://www.doombench.com/news/alibaba-releases-open-real-time-qwen2-5-omni-model-2025-03-27