Capability gains

Google introduces MUM for multimodal and multilingual search

Google introduced the Multitask Unified Model, a multimodal system trained across 75 languages and many tasks to address complex information needs; Google later deployed MUM features in Search worldwide.

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 DOOM43confidence 90/100

Why it moved the index

MUM expanded multilingual, multitask, and multimodal information-processing capability. Google's separately dated worldwide Search deployment verifies practical downstream impact, while the moving service identity and undisclosed exact checkpoint preclude a model attachment.

AUDIT TRAIL

Assessment history

  1. R1
    Toward 43 · confidence 90

    New May 2021 research result with separately verified practical Search deployment absent from the durable corpus.

    12 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Google introduces MUM for multimodal and multilingual search.
  1. DoomBench assesses “Google introduces MUM for multimodal and multilingual search” as evidence moving toward doom, with magnitude 43 and confidence 90 out of 100 in the capability gains category.

  2. The DoomBench assessment of “Google introduces MUM for multimodal and multilingual search” is based on reporting from Search Engine Land and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “Google introduces MUM for multimodal and multilingual search” as follows: Google introduced the Multitask Unified Model, a multimodal system trained across 75 languages and many tasks to address complex information...

    https://www.doombench.com/news/google-introduces-mum-for-multimodal-and-multilingual-search-2021-05-18