Congrats to @Alibaba_Qwen on releasing Qwen3.8-Flash-Next, an experimental open-weight model that pr...

TL;DR · AI 摘要
通义实验室发布Qwen3.8-Flash-Next模型,预览Qwen4架构,NVIDIA提供支持工具和API。
核心要点
- Qwen3.8-Flash-Next含125B参数和51B N-gram,预览Qwen4架构
- NVIDIA提供NeMo AutoModel和NeMo RL的Day 0微调支持
- Qwen3.8-Flash生产版将通过QwenCloud API提供,价格0.16美元/百万输入token
结构提纲
按章节快速跳转。
- §模型发布
通义实验室发布Qwen3.8-Flash-Next实验性开源模型
- ·技术特性
模型包含125B参数和51B N-gram,预览Qwen4架构
提供NeMo AutoModel和NeMo RL的Day 0微调支持
- ›集成方案
支持与sgl_project、vllm_project等工具链集成
生产版将通过QwenCloud API提供,定价0.16美元/百万输入token
思维导图
用一张图看清主题之间的关系。
查看大纲文本(无障碍 / 无 JS 友好)
- Qwen3.8-Flash-Next模型发布
- 模型特性
- 125B参数+51B N-gram
- Qwen4架构预览
- 支持工具
- NVIDIA NeMo AutoModel
- NVIDIA NeMo RL
- 商业化信息
- QwenCloud API定价方案
金句 / Highlights
值得收藏与分享的关键句。
Qwen3.8-Flash-Next含125B参数和51B N-gram,预览Qwen4架构
NVIDIA提供Day 0支持,可使用NeMo AutoModel和NeMo RL进行微调
Qwen3.8-Flash生产版定价0.16美元/百万输入token,0.47美元/百万输出token
NVIDIA AI on X: "Congrats to @Alibaba_Qwen on releasing Qwen3.8-Flash-Next, an experimental open-weight model that previews the Qwen4 architecture. We’ve got Day 0 support to fine-tune with NVIDIA NeMo AutoModel and NeMo RL, plus recipes to run it with @sgl_project, @vllm_project and" / X
NVIDIA AI
@NVIDIAAI
Congrats to
@
Alibaba_Qwen
on releasing Qwen3.8-Flash-Next, an experimental open-weight model that previews the Qwen4 architecture. We’ve got Day 0 support to fine-tune with NVIDIA NeMo AutoModel and NeMo RL, plus recipes to run it with
sgl_project
,
vllm_project
and
lightseekorg
TokenSpeed. Get started:
nvda.ws/4wV7yXp
Qwen
@Alibaba_Qwen
14h
⚡Meet Qwen3.8-Flash, a multimodal MoE and an early preview of the Qwen4 architecture, now open-weight! The production version Qwen3.8-Flash will be available soon via QwenCloud API at just $ 0.16/1M input tokens and $ 0.47/1M output tokens. 125B parameters + 51B N-gram
Show more
5:17 PM · Aug 26, 2026
24.8K
Views
17
32
368
40