AMD is excited to announce the integration of the new DeepSeek-V3 model from DeepSeek on AMD Instinct GPUs, optimized for performance powered by SGLang . This integration will help accelerate the development of cutting-edge AI applications and experiences. DeepSeek-V3 is an open-source, multimodal AI model designed to empower developers with unparalleled performance and efficiency. By seamlessly integrating advanced capabilities for processing both text and visual data, DeepSeek-V3 sets a new benchmark for productivity, driving innovation and enabling developers to create cutting-edge AI applications.

The DeepSeek-V3 model is a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were a part of its predecessor, DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. DeepSeek-V3 allows developers to work with advanced models, leveraging memory capabilities to enable processing text and visual data at once, enabling broad access to the latest advancements, and giving developers more features. DeepSeek-V3 achieves the best performance on most benchmarks, especially on math and code tasks.