QuixiAI Enables Cross-Backend Execution and Boosts CPU Inference

QuixiAI demonstrated a runtime covering six hardware backends including CUDA and CPU. Additionally, developers successfully ran BitNet on CPUs and achieved 100 tok/s inference speed for the Qwen3.6-35b model using two B60 XPUs.

2026-07-22 ~ 2026-07-22 · 3 related posts

• QuixiAI shows the same runtime spanning CUDA, Metal, ROCm, XPU, Gaudi and CPU — QuixiAI · 2026-07-22
• Qwen3.6-35b-a3b hits 100 tok/s on two B60 XPU cards with SYCL kernels — QuixiAI · 2026-07-22
• BitNet runs on CPU, while Qwen3.6-35B hits 100 tok/s on 2× B60 XPU — QuixiAI · 2026-07-22