Ox Alpha模型实为Zai的GLM-5.3-Flash
The viral "Ox Alpha" model is officially GLM-5.3-Flash from Zai. 🔍 📊 320B parameters with 18B act...
Zai的GLM-5.3-Flash模型就是之前爆火的Ox Alpha，320B参数但成本仅0.09美元，还用国产硬件跑通了。
Zai公司发布的GLM-5.3-Flash模型拥有320B参数，每 token 激活18B。该模型采用混合架构，结合线性和稀疏注意力机制。在GDPval AA v2基准测试中表现领先，单任务成本仅需0.09美元。Zai使用中国国产硬件完成了大规模免费预览，证明了智能内存优化可突破硬件限制。
The viral "Ox Alpha" model is officially GLM-5.3-Flash from Zai. 🔍 📊 320B parameters with 18B act...
The viral "Ox Alpha" model is officially GLM-5.3-Flash from Zai. 🔍 📊 320B parameters with 18B active per token ⚡ Hybrid architecture mixing linear and sparse attention 💰 Leading performance on real world tasks from GDPval AA v2 at just $0.09 per task The best part? Zai ran the massive free preview entirely on China-made hardware, proving smart memory optimization can overcome hardware limits. 🛠️ Dive into the architecture details in The Bat hubs.la/Q04wWZcM0 5e #DeepLearningAI n #LLMs L #OpenWeights ghts 💬 3 🔄 0 ❤️ 13 👀 1588 📊 4 ⚡
