What?! Alibaba has just released a new version of Qwen3.6... Which is better than Gemma 4 🤯 - Onl...
TL;DR · AI 摘要
阿里发布Qwen3.6-35B-A3B开源模型，仅3B激活参数，支持多模态与智能体编程，性能对标Claude Sonnet 4.5。
核心要点
- Qwen3.6-35B-A3B为稀疏MoE架构，总参35B但仅激活3B
- 采用Apache 2.0协议完全开源权重，支持商业使用
- 多模态能力与智能体编程表现突出，接近更大模型水平
Which is better than Gemma 4 🤯
- Only 3B active parameters
- Fully open weights (Apache 2.0)
- Multimodal (performance matches Claude Sonnet 4.5 )
And incredibly capable for agentic coding.
Super impressive! https://t.co/YF6F8z8UdA" / X
Paul Couvert on X: "What?! Alibaba has just released a new version of Qwen3.6... Which is better than Gemma 4 🤯 - Only 3B active parameters - Fully open weights (Apache 2.0) - Multimodal (performance matches Claude Sonnet 4.5 ) And incredibly capable for agentic coding. Super impressive! https://t.co/YF6F8z8UdA" / X
Don’t miss what’s happening
What?! Alibaba has just released a new version of Qwen3.6... Which is better than Gemma 4 
- Only 3B active parameters - Fully open weights (Apache 2.0) - Multimodal (performance matches Claude Sonnet 4.5 ) And incredibly capable for agentic coding. Super impressive!
Quote
Qwen
@Alibaba_Qwen
·
Apr 16
Meet Qwen3.6-35B-A3B：Now Open-Source！
A sparse MoE model, 35B total params, 3B active. Apache 2.0 license. 
Agentic coding on par with models 10x its active size 
Strong multimodal perception and reasoning ability 
Multimodal thinking + non-thinking modes
·
37
89
782
346
Read 37 replies