Intel Arc Pro B60 Professional GPU

The Intel Arc Pro B60 is a professional discrete GPU built on Intel's second-generation Xe2 (Battlemage) graphics architecture, offering a significant generational leap over its Alchemist predecessors in both compute throughput and memory capacity. At the heart of the card sits a full Battlemage Xe2 GPU die paired with 24GB of GDDR6 memory across a 192-bit memory interface, providing the large VRAM footprint that professional AI inference, large-model deployment, and complex visualization workloads demand. The Xe2 architecture introduces redesigned Xe-cores with improved INT8 and FP16 throughput, hardware-accelerated ray tracing, and enhanced XMX (Xe Matrix Extension) AI acceleration engines that dramatically increase tokens-per-second performance for on-device LLM inference compared to the previous generation.

The Arc Pro B60 is positioned squarely at professional workstation users who require certified driver support, ISV application compatibility, and the ability to run multiple GPUs in parallel for scaled-out inference or display-wall rendering. Intel's pro driver stack delivers stability certifications for leading DCC (digital content creation) applications, CAD tools, and AI frameworks including OpenVINO, PyTorch, and TensorFlow via Intel Extension for PyTorch (IPEX). The card supports up to four simultaneous display outputs and is designed to operate within standard workstation power envelopes, making it deployable in tower and rack-mount workstation form factors without requiring exotic cooling infrastructure.

What distinguishes the Arc Pro B60 in its competitive class is the combination of a substantial 24GB VRAM allocation — sufficient to load many 7B to 13B parameter quantized large language models entirely on a single card — with the hardware XMX matrix engines that accelerate INT4, INT8, and BF16 inference natively. For multi-GPU configurations, Intel provides unified software tooling that enables tensor parallelism across multiple B60 cards within a single workstation node, enabling professional studios and AI labs to scale inference capacity linearly. The card's launch in Q2 2025 marks Intel's maturation in the professional discrete GPU segment with a product designed from the ground up for the converging demands of generative AI and professional graphics.

Ideal for

• On-premises large language model inference for enterprise AI assistants, deploying quantized 7B–13B parameter models entirely within the 24GB VRAM footprint without cloud dependency
• Multi-GPU workstation nodes for tensor-parallel AI inference, combining two or more Arc Pro B60 cards to serve larger 30B+ parameter models with reduced latency
• Professional 3D visualization and real-time ray tracing in architectural visualization, product design, and media production pipelines using ISV-certified pro driver stacks
• GPU-accelerated video transcoding and AI-enhanced upscaling workflows in broadcast and post-production environments leveraging Xe Media Engine and XMX acceleration
• OpenVINO-optimized AI model deployment for computer vision inference at the workstation edge, including defect detection, medical imaging analysis, and geospatial processing
• Multi-display command-and-control visualization environments requiring up to four simultaneous high-resolution outputs from a single professional GPU in control rooms and simulation centers

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.