What if you could run the same code on AMD and NVIDIA with zero code changes?
With Modular Platform 25.4, that vision is now a reality! 
We’re proud to announce our official partnership with AMD, bringing full support for AMD Instinct™ MI300X and MI325X GPUs. You can now deploy the same container across both AMD and NVIDIA hardware with no code changes, no vendor lock-in, and no extra configuration!
Highlights from 25.4:
• Up to 53% better throughput on prefill-heavy BF16 workloads across Llama 3.1, Gemma 3, Mistral, and other state-of-the-art language models
• Support for AMD MI300/325, NVIDIA Blackwell, RTX 40xx, and RDNA3
• Expanded model support including Qwen3, OLMo2, Gemma3, and InternVL
• 450k+ lines of production-grade Mojo kernel code now open source
• Improved docs, PyTorch ops tutorials, and kernel performance tools
Get the full scoop in our release blog post: Modular: Modular 25.4: One Container, AMD and NVIDIA GPUs, No Lock-In
We’re also kicking off Modular Hack Weekend on June 27th with a GPU Programming Workshop and a stacked GPU prize pool! Join us virtually or in person: Modular Hack Weekend: GPU Programming Workshop · Luma