PulseAugur
EN
LIVE 12:37:02
中文(ZH) Nex-AGI 今天发布了 Nex-N2.6，321B 参数的多模态 Agent 模型，底子是 GLM-5.3-Flash，权重按 Apache 2.0 开源，FP8 量化版已经放上 HuggingFace 了！！ OSWorld 上拿了 83.0 分，比上一代的 82.2 高了 0.8 分，比底座 GLM-5.3-F

Nex-AGI releases open-source 321B multimodal agent model Nex-N2.6

By PulseAugur Editorial · [1 sources] ·

Nex-AGI has released Nex-N2.6, a 321B parameter multimodal agent model built on GLM-5.3-Flash. The model's weights are open-sourced under the Apache 2.0 license, with an FP8 quantized version available on Hugging Face. Nex-N2.6 achieved an OSWorld score of 83.0, an improvement over its predecessor, and significantly reduced repetition rates from 16.0% to 1.6%. AI

IMPACT Open-source release of a large multimodal agent model may accelerate research and development in agentic AI.

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

Read on Mastodon — mastodon.social →

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

Nex-AGI releases open-source 321B multimodal agent model Nex-N2.6

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, product
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 — mastodon.social TIER_1 中文(ZH) · Aven_NIxinhai ·

    Nex-AGI today released Nex-N2.6, a 321B parameter multimodal agent model, based on GLM-5.3-Flash, with weights open-sourced under Apache 2.0, and the FP8 quantized version is already on HuggingFace!! It scored 83.0 on OSWorld, 0.8 points higher than the previous generation's 82.2, and higher than the base GLM-5.3-F

    Nex-AGI 今天发布了 Nex-N2.6，321B 参数的多模态 Agent 模型，底子是 GLM-5.3-Flash，权重按 Apache 2.0 开源，FP8 量化版已经放上 HuggingFace 了！！ OSWorld 上拿了 83.0 分，比上一代的 82.2 高了 0.8 分，比底座 GLM-5.3-Flash 的 78.0 高了 5 分，DeepSeek-V4.1-Flash 只有 71.7。 重复推理率从上一代的 16.0% 砍到 1.6%，内部回放评测里相对降了 90%；工具列表遵守率直接拉到 100%，给什么工具就只用什么工具。 …

RELATED ENTITIES

RELATED TOPICS