readygpu

Intel Arc Pro B60 vs Intel Arc Pro B70

For running models locally, priced in Germany. At 32k context the first runs 57 models fully in memory, the second 61.

In short

The Intel Arc Pro B70 has more memory for models: 32 GB against 24 GB. At 32k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the Intel Arc Pro B60 57; only the Intel Arc Pro B70 holds Kimi Linear 48B-A3B. On the models both run fully in memory, the Intel Arc Pro B70 is about 1.3× as fast as the Intel Arc Pro B60 (estimated, median across models).

Updated . Speeds are estimates; how they are worked out.

Intel Arc Pro B60

953 €

Amazon

Add to build
Intel Arc Pro B70

Not priced here

Add to build

Specifications

SpecIntel Arc Pro B60Intel Arc Pro B70
GPUArc Pro B60Arc Pro B70
Memory24 GB GDDR632 GB GDDR6
Memory bus192-bit256-bit
Memory bandwidth456 GB/s608 GB/s
FP16 compute98.5 TFLOPS45.8 TFLOPS
SoftwareSYCL / oneAPISYCL / oneAPI
Board power200 W230 W
PCIe generationGen 5Gen 5
PCIe lanesx8x16

What each runs

Estimated decode speed with the most precise quant that fits each one est.

ModelIntel Arc Pro B60Intel Arc Pro B70
Qwen 3.8 27B10tok/sQ5_K_M10tok/sQ8_0
Qwen 3.6 27B11tok/sQ5_K_M10tok/sQ8_0
Gemma 4 31B10tok/sQ4_K_M11tok/sQ6_K
Qwen 3.6 35B-A3B48tok/sQ4_K_M50tok/sQ6_K
Qwen 3.5 27B10tok/sQ5_K_M10tok/sQ8_0
Gemma 4 26B-A4B28tok/sQ6_K33tok/sQ8_0
Qwen 3.5 35B-A3B49tok/sQ4_K_M53tok/sQ6_K
GLM 4.7 Flash24tok/sQ5_K_M29tok/sQ6_K
Devstral Small 2 24B10tok/sQ5_K_M10tok/sQ8_0
Nemotron 3.5 Lightning 30B-A3B51tok/sMXFP454tok/sUD-Q5_K_M
gpt-oss 20B39tok/sMXFP452tok/sMXFP4
Qwen3 Coder 30B-A3B25tok/sQ4_K_M29tok/sQ6_K
Kimi Linear 48B-A3Boffload78tok/sQ4_K_M
Qwen3 32Boffload10tok/sQ5_K_M
Gemma 4 12B17tok/sQ8_022tok/sQ8_0
Qwen 3.5 9B21tok/sQ8_029tok/sQ8_0
Mistral Small 3.2 24B10tok/sQ5_K_M10tok/sQ8_0
Gemma 3 27B10tok/sQ5_K_M10tok/sQ8_0
DeepSeek R1 Distill Qwen 32Boffload10tok/sQ5_K_M
Ministral 3 14B12tok/sQ8_015tok/sQ8_0
Qwen3 14B11tok/sQ8_014tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B17tok/sQ8_022tok/sQ8_0
Qwen3 8B17tok/sQ8_022tok/sQ8_0
Phi-4 14B10tok/sQ8_014tok/sQ8_0
Qwen 3.5 4B41tok/sQ8_054tok/sQ8_0
Ministral 3 8B17tok/sQ8_022tok/sQ8_0
Granite 4.2 8B16tok/sQ8_021tok/sQ8_0
Gemma 4 E4B25tok/sQ8_033tok/sQ8_0
LFM2.5 8B-A1B47tok/sQ8_063tok/sQ8_0
Llama 3.1 8B21tok/sQ6_K28tok/sQ6_K
Qwen2.5 0.5B244tok/sQ8_0326tok/sQ8_0
Qwen3.5 0.8B185tok/sQ8_0247tok/sQ8_0
Qwen3.5 2B123tok/sQ5_K_M164tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B81tok/sQ8_0107tok/sQ8_0
MiniCPM5 2B56tok/sQ8_074tok/sQ8_0
Spark X2.5 4B53tok/smlx-4bit71tok/smlx-4bit
LFM2.5 2.6B47tok/sQ8_062tok/sQ8_0
LFM2.5 VL 3B47tok/sQ8_062tok/sQ8_0
FrogNano 4B 260940tok/sQ8_053tok/sQ8_0
Mellum2.1 12B A2.5B Thinking40tok/sQ8_053tok/sQ8_0
granite vision 4.1 4b36tok/sQ8_048tok/sQ8_0
granite 4.2 3b35tok/sQ8_046tok/sQ8_0
Ministral 3 3B Instruct 251232tok/sQ8_043tok/sQ8_0
Kimi VL A3B Thinking 250630tok/sQ8_041tok/sQ8_0
Kimi VL A3B29tok/smlx-8bit39tok/smlx-8bit
LFM2.5 24B A2B28tok/sQ6_K35tok/sQ8_0
GLM 4 9B 041424tok/smlx-5bit32tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B23tok/sQ8_031tok/sQ8_0
GLM 4.6V Flash20tok/sQ8_027tok/sQ8_0
DeepSeek R1 Distill Llama 8B18tok/sQ8_024tok/sQ8_0
GLM 4.1V 9B Thinking17tok/smlx-8bit23tok/smlx-8bit
Ministral 3 8B Reasoning 251217tok/sQ8_022tok/sQ8_0
Ministral 3 14B Reasoning 251212tok/sQ8_015tok/sQ8_0
Muse Glimmer 30B11tok/sQ5_K_M13tok/sQ6_K
DeepSeek R1 Distill Qwen 14B10tok/sQ8_014tok/sQ8_0
Devstral Small 250710tok/sQ5_K_M10tok/sQ8_0
Magistral Small 250610tok/sQ5_K_M10tok/sQ8_0
Magistral Small 250710tok/sQ5_K_M10tok/sQ8_0
Magistral Small 250910tok/sQ5_K_M10tok/sQ8_0
Qwen3.8 27B OBLITERATED10tok/smlx-4bit13tok/smlx-4bit
granite 4.2 30boffload10tok/sQ5_K_M
Qwen 3.5 122B-A10Boffloadoffload
Qwen3 Coder Nextoffloadoffload
Nemotron 3 Super 120B-A12Bwon't runoffload
Mistral Small 4 119Boffloadoffload
GLM 4.5 Airoffloadoffload
Qwen3 Next 80B-A3Boffloadoffload
Llama 3.3 70Bwon't runoffload
DeepSeek R1 Distill Llama 70Bwon't runoffload
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
Qwen 3.8 Flash Nextwon't runwon't run
MiniMax M2.7won't runwon't run
gpt-oss 120Bwon't runwon't run
DeepSeek R1won't runwon't run
DeepSeek R1 0528won't runwon't run
DeepSeek R1 Zerowon't runwon't run
DeepSeek V3won't runwon't run
DeepSeek V3 0324won't runwon't run
DeepSeek V3.1won't runwon't run
DeepSeek V3.1 Terminuswon't runwon't run
DeepSeek V3.2won't runwon't run
DeepSeek V4 Flash 0731won't runwon't run
DeepSeek V4 Flash Vision Expwon't runwon't run
Devstral 2 123B Instruct 2512won't runwon't run
GLM 4.5won't runwon't run
GLM 4.5Vwon't runwon't run
GLM 4.6won't runwon't run
GLM 4.6Vwon't runwon't run
GLM 4.7won't runwon't run
GLM 5won't runwon't run
GLM 5.1won't runwon't run
GLM 5.2won't runwon't run
GLM 5.3won't runwon't run
Kimi Dev 72Bwon't runwon't run
Kimi K2won't runwon't run
Kimi K2 Instruct 0905won't runwon't run
Kimi K2 Thinkingwon't runwon't run
Kimi K2.5won't runwon't run
MiniMax M2won't runwon't run
MiniMax M2.1won't runwon't run
MiniMax M2.5won't runwon't run
MiniMax M3won't runwon't run
Mistral Medium 3.5 128Bwon't runwon't run

Questions

Which is better for local AI, the Intel Arc Pro B60 or the Intel Arc Pro B70?
Choose the Intel Arc Pro B70 to run bigger models. Choose the Intel Arc Pro B70 for speed.
Which can run bigger models, the Intel Arc Pro B60 or the Intel Arc Pro B70?
At 32k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the Intel Arc Pro B60 57; only the Intel Arc Pro B70 holds Kimi Linear 48B-A3B.
Which is faster for local AI, the Intel Arc Pro B60 or the Intel Arc Pro B70?
On the models both run fully in memory, the Intel Arc Pro B70 is about 1.3× as fast as the Intel Arc Pro B60 (estimated, median across models).

More comparisons

More hardware for local AI