Intel Arc Pro B70 Benchmarks With LLM / AI, OpenCL, OpenGL & Vulkan

Written by Michael Larabel in Graphics Cards on 8 April 2026 at 11:28 AM EDT. Page 3 of 11. 65 Comments.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Qwen3-8B-Q8_0, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: gpt-oss-20b-Q8_0, Test: Text Generation 128. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: gpt-oss-20b-Q8_0, Test: Prompt Processing 512. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: gpt-oss-20b-Q8_0, Test: Prompt Processing 1024. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: gpt-oss-20b-Q8_0, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: GLM-4.7-Flash-IQ4_XS, Test: Text Generation 128. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: GLM-4.7-Flash-IQ4_XS, Test: Prompt Processing 512. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: GLM-4.7-Flash-IQ4_XS, Test: Prompt Processing 1024. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: GLM-4.7-Flash-IQ4_XS, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 512. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Mistral-7B-Instruct-v0.3-Q8_0, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

For those interested in Vulkan for AI or fans of Llama.cpp, I did run some Llama.cpp Vulkan benchmarks as part of this initial assortment of benchmarks. Compared to the other Battlemage and Alchemist graphics cards tested on Ubuntu 26.04, the Arc Pro B70 was delivering very strong performance as expected.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: granite-3.0-3b-a800m-instruct-Q8_0, Test: Text Generation 128. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: DeepSeek-R1-Distill-Llama-8B-Q8_0, Test: Prompt Processing 512. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: DeepSeek-R1-Distill-Llama-8B-Q8_0, Test: Prompt Processing 1024. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: granite-3.0-3b-a800m-instruct-Q8_0, Test: Prompt Processing 512. Arc Pro B70 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: granite-3.0-3b-a800m-instruct-Q8_0, Test: Prompt Processing 2048. Arc Pro B70 was the fastest.

It's great seeing Llama.cpp continuing to work out well atop the Vulkan API. The Arc Pro B70 performance was strong and in the coming days we'll see how it stacks up against the AMD and NVIDIA cards for testing. Looking ahead, as covered yesterday is also the new Intel "Jay" compiler coming about for the Intel ANV Vulkan and Iris drivers. This Jay compiler may help provide for better Vulkan performance in time.

65 Comments - Next Page

Page: 1 2 3 4 5 6 7 8 9 10 11 Next Page

URL: https://www.phoronix.com/review/intel-arc-pro-b70-linux/3

⇱ Intel Arc Pro B70 Benchmarks With LLM / AI, OpenCL, OpenGL & Vulkan Review - Phoronix

Intel Arc Pro B70 Benchmarks With LLM / AI, OpenCL, OpenGL & Vulkan