PulseAugur
EN
LIVE 00:24:41
한국어(KO) Qwen팀이 Qwen3.8-Flash-Next 가중치를 공개했습니다. 멀티모달 MoE 모델로 Gated DeltaNet+Qwen Sparse Attention(GDN+QSA) 하이브리드, Gated Residual, N-gram 임베딩, Muon 옵티마이저 등을 도입해 성능은 올리고 연

Qwen releases Qwen3.8-Flash-Next multimodal MoE model with 1M context

By PulseAugur Editorial · [1 sources] ·

The Qwen team has released the weights for their Qwen3.8-Flash-Next model, a multimodal Mixture-of-Experts (MoE) architecture. This new model incorporates innovations such as Gated DeltaNet+Qwen Sparse Attention (GDN+QSA) hybrid, Gated Residual Networks, N-gram embeddings, and the Muon optimizer to enhance performance while reducing computational and training costs. The model supports a base context of 262,144 tokens, expandable to 1 million tokens via YaRN, and its weights are available on Hugging Face and ModelScope. AI

IMPACT This release offers enhanced efficiency and a significantly larger context window, potentially enabling more complex multimodal applications.

RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Qwen releases Qwen3.8-Flash-Next multimodal MoE model with 1M context

How we ranked this

Signal score
19 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Significant
Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. Mastodon — fosstodon.org TIER_1 한국어(KO) · [email protected] ·

    Qwen team released Qwen3.8-Flash-Next weights. It introduces a multimodal MoE model with Gated DeltaNet+Qwen Sparse Attention (GDN+QSA) hybrid, Gated Residual, N-gram embedding, and Muon optimizer to improve performance and reduce computation.

    Qwen팀이 Qwen3.8-Flash-Next 가중치를 공개했습니다. 멀티모달 MoE 모델로 Gated DeltaNet+Qwen Sparse Attention(GDN+QSA) 하이브리드, Gated Residual, N-gram 임베딩, Muon 옵티마이저 등을 도입해 성능은 올리고 연산·학습 비용은 크게 절감했습니다. 핵심 모델은 125B 파라미터(추가 51B N-gram), 기본 262,144토큰 지원·YaRN으로 1M 확장 가능. 가중치는 Hugging Face·ModelScope에 공개되며…

    LINKS qwen.ai/blog qwen.ai/blog

RELATED ENTITIES

RELATED TOPICS