Latest news and articles about inference cost
Total: 1 articles found
Alibaba’s Qianwen has open‑sourced Qwen3‑Coder‑Next, an 80B parameter model designed for coding agents and local deployment that combines hybrid attention with MoE to lower inference costs. The release aims to accelerate enterprise adoption in China by enabling on‑premise use and customization, while raising questions about IP, safety and the infrastructure needed to realize claimed efficiency gains.
