Tether: QVAC SDK 0.15.0 Adds AMD GPU Backend

Tether CEO Paolo Ardoino announced the QVAC SDK 0.15.0 release today, delivering prompt batching for concurrent LLM jobs, a native HIP/ROCm AMD GPU backend that outperforms Vulkan by 23 percent, and OpenClaw as the second local coding-agent integration alongside OpenCode. Vision encoders for Qwen3.5 and Gemma-4 now run on flagship Android GPUs via Vulkan and OpenCL, while transcription and text-to-speech pipelines gain faster CPU fallbacks and up to 48 kHz output. The update strengthens Tether AI infrastructure for fully local, cloud-free workflows across Linux and mobile platforms.