AI news story

Researchers train AI model that hits near-full performance with just 12.5 percent of its experts

Researchers at the Allen Institute for AI and UC Berkeley have built EMO, a mixture-of-experts model whose experts specialize in content domains instead of word types. That lets you strip out three-quarters of the experts while losing only about one

  • AI
  • Source: The Decoder
  • Published: 2026-05-16
  • Signal score: 4
  • 14 sources

Editor's take

A new research paper details an AI model, EMO, that achieves near-peak performance with significantly fewer specialized components. This "mixture-of-experts" approach, where experts focus on content domains rather than linguistic units, demonstrates that a substantial portion of the model's knowledge can be pruned without a proportional drop in efficacy.

This development is significant as it addresses the escalating computational demands and costs associated with large AI models. By proving that efficient specialization can lead to robust performance even when parts are removed, it opens avenues for more accessible and deployable AI systems, potentially impacting how companies like OpenAI or Google design and scale their future foundational models.

Future research should focus on the specific content domains EMO excels in and whether this domain-based expert specialization translates to other modalities like image or video generation. Understanding the exact trade-offs in performance and the potential for further optimization will be critical in determining the practical applicability of this efficiency gain.

Signal score: 4

This event was corroborated by 14 independent sources. The signal score weighs cross-source corroboration, recency, source weight and topic salience. How we rank stories.

More AI stories

  1. Meet Shepherd: An Open-Source Python Substrate That Lets Meta-Agents Fork, Replay, and Revert Any Agent Run

    MarkTechPost · 2026-08-08

    Long agent runs accumulate state that no transcript records — edited files, a live dev server, installed packages, a warm prompt cache.

  2. Denmark Requires Oral Defenses for Students' Written Work to Counter AI Cheating

    Hacker News · 2026-08-08

    Denmark's Ministry of Education has mandated oral defenses for student assignments to mitigate AI-generated content.

  3. Cloudflare launches Kitesurf, a browser built for AI agents

    TechCrunch · 2026-08-07

    Kitesurf is a cloud-hosted browser designed for AI agents instead of people. It uses less computing power than Chromium for common automation tasks

  4. Pokee AI Releases Pokee-Isaac 28B: A 10M-Token Context Agentic Model Built to Run Inside the Customer Boundary

    MarkTechPost · 2026-08-08

    Pokee AI released Pokee-Isaac 28B, a 28B text-only foundation model with a 10M-token context window built to run inside the customer boundary.

  5. Gentoo bugzilla closed due AI bot scraper overload

    Hacker News · 2026-08-08

    The Gentoo Bugzilla instance has been taken offline due to an overwhelming volume of automated traffic from an AI model scraper.

  6. Before Q, K, and V: Reconstructing the Transformer

    Towards Data Science · 2026-08-08

    Many Transformer explainers start with the finished architecture. We ask why it looks the way it does.