Gemma 4 Gets Day 0 Support on AMD GPUs and AI CPUs

AMD has announced Day 0 support for Google’s Gemma 4 across its entire hardware ecosystem, signaling a major push into open AI deployment.

The company confirmed that Gemma 4 runs across Radeon GPUs, Instinct accelerators, and Ryzen AI CPUs, enabling a unified experience from consumer PCs to datacenter infrastructure.

Full-stack support across AMD hardware

AMD designed this rollout to cover its full product stack. Gemma 4 works on Radeon GPUs used in gaming and professional systems, Instinct GPUs built for datacenter workloads, and Ryzen AI CPUs that power AI PCs with dedicated NPUs.

This unified approach allows developers to build once and deploy anywhere without switching platforms or rewriting workflows.

Strong ecosystem support from day one

AMD paired Gemma 4 with broad software compatibility across popular AI tools. The model works with frameworks like vLLM and SGLang for high-performance inference, while also supporting lightweight environments such as llama.cpp.

Local AI applications like LM Studio and Ollama also integrate seamlessly, and developers can access OpenAI-style APIs through Lemonade Server. AMD offers flexible deployment options through Docker images, Python packages, or direct local execution.

Built for performance and scale

AMD emphasized performance as a key advantage of its implementation. The company optimized Gemma 4 for concurrent inference workloads using vLLM, while SGLang enables high-throughput serving in production environments.

One of the standout claims is the ability to run the full Gemma 4 model on a single MI300X GPU with 192GB of VRAM, reducing the need for multi-GPU setups in certain scenarios.

Hardware-specific acceleration also plays a major role. Radeon GPUs benefit from ROCm optimizations, while Ryzen AI CPUs leverage XDNA 2 NPUs to accelerate on-device AI processing.

More optimizations coming soon

AMD confirmed that this is just the beginning. Future updates will bring additional optimizations for MI300 and upcoming MI350 GPUs, further improving performance in enterprise environments.

The company also plans to expand NPU support for smaller Gemma models such as E2B and E4B, which will improve efficiency on AI PCs and edge devices.

AMD pushes deeper into open AI

With Day 0 support for Gemma 4, AMD continues to position itself as a full-stack AI platform that spans consumer hardware, enterprise infrastructure, and local AI applications.

The company’s focus on open-source tools and flexible deployment highlights its strategy to compete across both developer ecosystems and large-scale AI deployments.

In other news, AMD recently launched FSR 4.1, while Sony confirmed that its PSSR technology shares similarities with AMD’s approach. AMD has also rebranded Anti-Lag as FSR Latency Reduction 2.0 as part of its broader ecosystem push.

Via Wccftech