Intel Arc Pro B60 24GB Specs, Benchmarks & Pricing
The Intel Arc Pro B60 is a professional workstation GPU released in September 2025, built on the Xe2-HPG (Battlemage) architecture using TSMC's 5nm process. It features 20 Xe2 cores, 160 XMX (Xe Matrix eXtension) engines for AI acceleration, and 20 ray tracing units, paired with 24GB of GDDR6 memory. The GPU delivers 12.28 TFLOPS of FP32 compute performance and 197 peak TOPS of INT8 AI inference performance, making it highly competitive for AI workstation and inference workloads at its $599 price point. The XMX engines support precision formats including INT4, INT8, FP16, and BF16, providing exceptional flexibility for AI inference tasks that can leverage lower precision for improved performance. The Arc Pro B60 features 456 GB/s of memory bandwidth through a 192-bit memory interface running at 19 Gbps, with a boost clock of 2400 MHz and a TBP of 200W (with a low-power mode available at 120-200W and 2000 MHz). Intel positions this GPU as an 'inference workstation GPU' that is 'LLM inference ready', differentiating it from the gaming-focused Arc B580 by doubling the memory capacity from 12GB to 24GB. The GPU supports PCIe 5.0 x8 interface and features 4 DisplayPort 2.1 outputs (one supporting UHBR 13.5, three supporting UHBR 10) and HDMI 2.1, with support for up to four displays at 7680x4320 resolution. The card has been tested and certified for professional applications including Autodesk Maya, SolidWorks, Adobe Premiere Pro, Blender, and DaVinci Resolve. The GPU also supports SR-IOV, Linux containers, and parallel multi-GPU operation, with Intel's 'Project Battlematrix' demonstrating systems with up to 8 Arc Pro B60 GPUs providing 192GB of combined VRAM for demanding AI inference workloads.
- Release Date: September 19, 2025
- MSRP: $599 USD
- GPU Architecture: Xe2-HPG (Battlemage)
- Hardware-Accelerated GEMM Operations:FP64 FP32 TF32 BF16 FP16 FP8 FP4 INT8 INT4
- CUDA Compute Capability : n/a
Considerations
- Lower FP32 compute performance compared to other GPUs
- Lower FP16 compute performance compared to other GPUs
Specifications for Intel Arc Pro B60
| Specification | Performance Ranking |
|---|---|
| FP32 TFLOPs | |
| FP16 TFLOPs | |
| Tensor Core Count | |
| Memory Capacity (GB) | |
| Memory Bandwidth (GB/s) | |
| Int8 TOPs | |
| FP8 TFLOPs | Data not available |
| FP4 TFLOPs | Data not available |
Real-time Intel Arc Pro B60 GPU Prices
Compare Price/Performance to other GPUs
Compare Intel Arc Pro B60 to Another GPU
Price History
Intel Arc Pro B60 Price History
This chart tracks Intel Arc Pro B60 prices over the past 6 months based on eBay listings. The yellow line shows the average of the three lowest-priced listings each day—a useful indicator for finding deals.
Trend: Over this period, the lowest average price has been down 4.1%. The current lowest average is $650, compared to a typical $675 over the period. The lowest average price is calculated as the average of the three lowest-priced listings each day.
Product Identifiers
Manufacturer Part Numbers (1)
- Product Code
- 243916
Available from 2 Partners (3 products)
- ASRock Intel Arc Pro B60 Creator 24GB
- ARC PRO B60 CREATOR 24GB(model number)
- ASRock Intel Arc Pro B60 Passive 24GB
- ARC PRO B60 PASSIVE 24GB(model number)
- Maxsun Intel Arc Pro B60 Dual 48G Turbo
- ARC PRO B60 DUAL 48G TURBO(model number)
References
- https://www.intel.com/content/www/us/en/products/sku/243916/intel-arc-pro-b60-graphics/specifications.html
- https://www.intel.com/content/www/us/en/products/docs/discrete-gpus/arc/workstations/b-series/overview.html
- https://download.intel.com/newsroom/2025/client-computing/Intel-Arc-Pro-B60-Data-Sheet.pdf
- https://newsroom.intel.com/client-computing/computex-intel-unveils-new-gpus-ai-workstations
- https://wccftech.com/intel-arc-pro-b60-24-gb-b50-16-gb-battlemage-gpus-pro-ai-3x-faster-dual-gpu-variant/
- https://www.tomshardware.com/pc-components/gpus/intel-launches-usd299-arc-pro-b50-with-16gb-of-memory-project-battlematrix-workstations-with-24gb-arc-pro-b60-gpus
- https://www.asrock.com/Graphics-Card/Intel/Intel%20Arc%20Pro%20B60%20Creator%2024GB/
- https://asrock.com/Graphics-Card/Intel/Intel%20Arc%20Pro%20B60%20Passive%2024GB/
- https://www.maxsun.com/products/intel-arc-pro-b60-dual-48g-turbo
- https://siliconcat.net/gpu/intel-arc-pro-b60
- https://cputronic.com/gpu/intel-arc-pro-b60
- https://www.hwcooling.net/en/batttlemage-details-of-intel-xe2-gpu-architecture-analysis/
- https://chipsandcheese.com/p/lunar-lakes-igpu-debut-of-intels
- https://en.wikipedia.org/wiki/Intel_Arc
- https://community.intel.com/t5/GPU-Compute-Software/What-is-the-Intel-Arc-pro-b60-quot-FP64-quot-performance/m-p/1722047
- https://github.com/intel/compute-runtime/blob/master/shared/source/dll/devices/devices_base.inl
- https://www.intel.com/content/www/us/en/docs/vtune-profiler/user-guide/2025-1/xve-fp16-bf16-int8-int4-int2-xmx-instructions.html
- https://www.intel.com/content/dam/www/central-libraries/us/en/documents/2026-03/datasheet-b70-gpu.pdf
Notes
- FP32 TFLOPS of 12.28 sourced directly from Intel's official product specifications page at intel.com.
- FP16 TFLOPS of 24.58 sourced from third-party GPU databases (SiliconCat, CpuTronic) based on Xe2 architecture specifications. Intel does not publish FP16 performance in official specs, but the 2:1 ratio of FP16 to FP32 is consistent with Xe2 architecture documentation.
- INT8 TOPS of 197 represents peak INT8 AI performance (Int8 Dense per Intel's official B60 datasheet) and is sourced directly from Intel's official product specifications.
- Memory bandwidth calculated as 192-bit interface × 19 Gbps ÷ 8 bits/byte = 456 GB/s, confirmed by Intel official specifications.
- TBP (Thermal Base Power) is 200W at 2400 MHz boost clock. Intel also specifies a low-power mode ranging from 120-200W at 2000 MHz.
- XMX (Xe Matrix eXtension) engine count of 160 sourced from Intel official specifications. These are Intel's dedicated AI accelerators, equivalent to tensor cores in NVIDIA GPUs.
- FP64 double precision performance is not officially published by Intel. Third-party sources report FP64 at 1:4 ratio to FP32 (approximately 3.07 TFLOPS), but this is unverified. FP64, when supported, runs on the general-purpose shader units (XVEs), not the XMX matrix engine, so it is not included in supportedHardwareOperations under the GEMM-only definition (see Phase 5 audit note below).
- The GPU uses the BMG-G21 chip with 272mm² die size and 19.6 billion transistors, manufactured on TSMC's N5 (5nm) process.
- MSRP of $599 USD based on retail pricing at launch from US retailer Central Computers for the ASRock Arc Pro B60 Creator 24GB. Intel did not officially publish MSRP but indicated an estimated value around $500 at Computex 2025 announcement.
- Release date of September 19, 2025 based on multiple sources confirming the GPU launched in September 2025 with availability in North America. The GPU was announced at Computex 2025 in May 2025, with sampling to AIB partners beginning in June 2025.
- Intel Product SKU 243916 sourced from Intel's official product page URL structure.
- Device ID 0xE211 sourced from Intel's official specifications page and confirmed in Intel's compute-runtime GitHub repository.
- AIB partners include ASRock, Gunnir, Lanner, Maxsun, Onix, Senao, and Sparkle. Specific part numbers for individual AIB products are not widely published yet, as most products were listed with model names only.
- The GPU shares its silicon foundation (BMG-G21) with the gaming-focused Arc B580, both using the same die with 20 Xe2 cores. The critical differentiation is memory capacity (24GB vs 12GB) and market positioning.
- Intel positions the Arc Pro B60 as delivering up to 3× faster performance compared to previous generation professional GPUs at this price point, with significantly better performance per dollar.
- The GPU supports DirectX 12 Ultimate and OpenGL 4.6 APIs.
- Intel's Project Battlematrix showcases dual-GPU and multi-GPU configurations, with systems featuring up to 8 Arc Pro B60 GPUs providing 192GB of combined VRAM for large-scale AI inference workloads.
- Phase 5 precision audit (2026-09-02): supportedHardwareOperations corrected to FP16, BF16, INT8, INT4 (removed FP32, FP64, and TF32) per the canonical Xe2-HPG (Battlemage) supportedHardwareOperations definition in gpu-researcher.md. TF32 was a data-entry inconsistency: it is not listed anywhere in Intel's official Arc Pro B60 datasheet or specifications page, is not present on the sibling B570/B580/Pro B50 files (same Xe2-HPG generation, same XMX design), and appears to have been over-applied from Intel's generic oneAPI XMX documentation ("Intel XMX supports...int8, fp16, bf16, and tf32...depending on hardware generation") which describes TF32 support on the separate Xe-HPC (Ponte Vecchio / Data Center GPU Max Series) architecture, not Xe2/Battlemage. FP32 and FP64 were removed because this field is scoped to hardware-accelerated XMX matrix/tensor (GEMM) paths only; FP32/FP64 run on the general-purpose Xe Vector Engines (XVEs), per Intel VTune Profiler documentation which defines a dedicated XMX instruction-count metric scoped to FP16/BF16/INT8/INT4/INT2 only, distinct from a separate XVE metric covering INT16/INT32/INT64/FP16/FP32/FP64.
- fp8TFLOPS and fp4TFLOPS are intentionally omitted (not added to unverifiedSpecs) because this is a confirmed hardware negative, not a research gap: Intel's official Arc Pro B60 datasheet (which does list "Int8 Dense" AI TOPS explicitly) and the B70 datasheet make no mention of FP8 or FP4 XMX data type support anywhere, and the chipsandcheese Battlemage architecture analysis likewise documents no FP8/FP4 XMX support for the Xe2/Xe2-HPG generation.
- gpuArchitecture changed from "Xe2" to "Xe2-HPG (Battlemage)" during the 2026-09-02 audit to normalize naming across all six Arc B/Pro B-series files. Intel's own official Arc Pro B60 datasheet (PDF one-pager, download.intel.com/newsroom/2025/client-computing/Intel-Arc-Pro-B60-Data-Sheet.pdf) explicitly labels its Xe-cores "20 Xe2-HPG" in the Specifications table, confirming Intel itself uses the "Xe2-HPG" designation for this Pro-tier product, not just consumer B570/B580. No genuine hardware-based Pro-vs-consumer naming split was found; both tiers share the same Xe2-HPG/Battlemage generation and BMG-G21 die.