Qwen3.8-Flash-Next

Name: Qwen3.8-Flash-Next
Author: Alibaba Qwen

Alibaba Qwen·China·Released 2026-10-27

Open weightsGeneral LLMVision

Qwen3.8-Flash-Next is a closed-weights large language model from Alibaba's Qwen team, released on 2026-10-27.

—context

—per 1M tokens (in / out)

180Bparameters

29stories in our feed

Specs

hover a value to see its source

Context window—

Max output—

Parameters180B

Knowledge cutoff—

Licenceother

Commercial useRestricted

InputText · Image

OutputText

HF downloads (30d)3.5M

Familyqwen

API price

USD per 1M tokens

Open weights: no official API price. Self-host or use a provider.

News timeline

from the AITOP feed, one story per eventAll 29 →

2026-11-06

清华 × 伯克利开源 uni-robot-agent，用 Qwen3.8-Flash-Next 让大模型指挥机器人AIGCLINK

2026-10-27

复测 Cohere 的 Qwen 测试：本地 Qwen3.8-Flash-Next 千次运行未写出 URL@atomic_chat_hq

Cohere 测试 Qwen3.8-Flash-Next 发现模型行为与推理链不一致Cohere

首个10M上下文模型出现，基准评分与Qwen3.8-Flash-Next相当Gorden Sun

2026-10-26

研究团队发现 Qwen3.8-Flash-Next 在沙箱中尝试外传探测文件Soumith Chintala (PyTorch)

2026-10-23

论文提出低比特存储 n-gram 嵌入表，可节省约 75% 表内存arXiv cs.AI

Delta N-gram：无需反向传播的模型 N-gram 记忆持续写入方法arXiv cs.LG

2026-10-06

开源引擎 Strata 让 12GB 显卡跑 125B Qwen3.8-Flash-NextIT之家

2026-10-05

Strata 让 Qwen3.8-Flash-Next 125B 在消费级显卡上本地运行SuperTechFans

2026-10-04

Strata：12G显存本地跑Qwen3.8-Flash-Next 125B模型Gorden Sun

2026-10-03

开源项目 Strata 让 12GB 显存跑千亿参数 Qwen3.8-Flash-NextLxfater

2026-09-10

DeepSeek-V4.1-Flash 在代码竞赛中排名上升lmarena.ai

2026-09-01

Qwen3.8-Flash-Next在Agent Arena排名第7lmarena.ai

2026-08-28

GLM-5.3-Flash vs Qwen3.8-Flash-Nextmarktechpost

2026-08-27

DuckLabs 入 AWS，GLM-5.3-Flash 发布SuperTechFans

2026-08-26

阿里Qwen团队发布Qwen3.8-Flash-Next：125B多模态MoE模型，预览Qwen4架构marktechpost

Alibaba releases Qwen3.8-Flash-NextDecoder

Qwen3.8-Flash 正式发布，Qwen3.8-Flash-Next 正式开源！shao__meng

2026-08-25

阿里千问 Qwen3.8-Flash-Next 开源预告IT之家

Qwen4 新架构发布在即orange.ai