readygpu

Intel Arc Pro B70 vs Mac mini M4 32GB

For running models locally, priced in the UK. At 8k context the first runs 61 models fully in memory, the second 58.

In short

Both have 32 GB of memory for models; the Intel Arc Pro B70 has 608 GB/s of bandwidth and the Mac mini M4 32GB 120 GB/s. At 8k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the Mac mini M4 32GB 58; only the Intel Arc Pro B70 holds Qwen 3.5 35B-A3B. On the models both run fully in memory, the Intel Arc Pro B70 is about 3.9× as fast as the Mac mini M4 32GB (estimated, median across models).

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

Intel Arc Pro B70

Not priced here

Add to build
Mac mini M4 32GB

£1,599

eBay

Add to build

Specifications

SpecIntel Arc Pro B70Mac mini M4 32GB
GPUArc Pro B70–
Memory32 GB GDDR6–
Memory bus256-bit128-bit
Memory bandwidth608 GB/s120 GB/s
FP16 compute45.8 TFLOPS3.76 TFLOPS
SoftwareSYCL / oneAPIMetal
Board power230 W–
PCIe generationGen 5–
PCIe lanesx16–
Chip–Apple M4 (10-core GPU)
Unified memory–32 GB
GPU can use by default–21.3 GB
GPU can use, tuned–24 GB
Operating system–macos
Processor–M4 10-core
Storage–512 GB

What each runs

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

ModelIntel Arc Pro B70Mac mini M4 32GB
Qwen 3.8 27B10tok/sQ8_04tok/sQ5_K_M
Qwen 3.6 27B10tok/sQ8_04tok/sQ5_K_M
Gemma 4 31B11tok/sQ6_K4tok/sQ4_K_M
Qwen 3.6 35B-A3B59tok/sQ6_K23tok/sMXFP4
Qwen 3.5 27B10tok/sQ8_04tok/sQ5_K_M
Gemma 4 26B-A4B37tok/sQ8_013tok/sUD-Q5_K_M
Qwen 3.5 35B-A3B63tok/sQ6_Kwon't run
GLM 4.7 Flash29tok/sQ8_012tok/sQ5_K_M
Devstral Small 2 24B12tok/sQ8_04tok/sQ6_K
Nemotron 3.5 Lightning 30B-A3B56tok/sUD-Q5_K_Mwon't run
Kimi Linear 48B-A3B86tok/sQ4_K_Mwon't run
gpt-oss 20B64tok/sMXFP418tok/sMXFP4
Qwen3 Coder 30B-A3B51tok/sQ6_K18tok/sQ4_K_M
Qwen3 32B11tok/sQ6_K4tok/sQ4_K_M
Gemma 4 12B23tok/sQ8_06tok/sQ8_0
Qwen 3.5 9B31tok/sQ8_08tok/sQ8_0
Mistral Small 3.2 24B12tok/sQ8_04tok/sQ6_K
DeepSeek R1 Distill Qwen 32B11tok/sQ6_K4tok/sQ4_K_M
Gemma 3 27B10tok/sQ8_04tok/sQ5_K_M
Ministral 3 14B19tok/sQ8_05tok/sQ8_0
Qwen3 14B18tok/sQ8_05tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B31tok/sQ8_08tok/sQ8_0
Qwen3 8B31tok/sQ8_08tok/sQ8_0
Phi-4 14B18tok/sQ8_05tok/sQ8_0
Qwen 3.5 4B63tok/sQ8_016tok/sQ8_0
Ministral 3 8B30tok/sQ8_08tok/sQ8_0
Granite 4.2 8B28tok/sQ8_07tok/sQ8_0
Gemma 4 E4B36tok/sQ8_09tok/sQ8_0
LFM2.5 8B-A1B114tok/sQ8_032tok/sQ8_0
Llama 3.1 8B40tok/sQ6_K10tok/sQ6_K
Qwen2.5 0.5B481tok/sQ8_0124tok/sQ8_0
Qwen3.5 0.8B326tok/sQ8_084tok/sQ8_0
Qwen3.5 2B196tok/sQ5_K_M50tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B143tok/sQ8_037tok/sQ8_0
MiniCPM5 2B100tok/sQ8_026tok/sQ8_0
LFM2.5 2.6B90tok/sQ8_023tok/sQ8_0
LFM2.5 VL 3B90tok/sQ8_023tok/sQ8_0
Spark X2.5 4B89tok/smlx-4bit23tok/smlx-4bit
granite vision 4.1 4b71tok/sQ8_018tok/sQ8_0
Ministral 3 3B Instruct 251267tok/sQ8_017tok/sQ8_0
granite 4.2 3b67tok/sQ8_017tok/sQ8_0
FrogNano 4B 260962tok/sQ8_016tok/sQ8_0
LFM2.5 24B A2B60tok/sQ8_019tok/sQ6_K
Mellum2.1 12B A2.5B Thinking59tok/sQ8_017tok/sQ8_0
Kimi VL A3B Thinking 250650tok/sQ8_014tok/sQ8_0
Kimi VL A3B48tok/smlx-8bit13tok/smlx-8bit
DeepSeek R1 Distill Qwen 7B36tok/sQ8_09tok/sQ8_0
GLM 4 9B 041435tok/smlx-5bit9tok/smlx-5bit
DeepSeek R1 Distill Llama 8B32tok/sQ8_08tok/sQ8_0
Ministral 3 8B Reasoning 251230tok/sQ8_08tok/sQ8_0
GLM 4.6V Flash29tok/sQ8_08tok/sQ8_0
GLM 4.1V 9B Thinking25tok/smlx-8bit6tok/smlx-8bit
Ministral 3 14B Reasoning 251219tok/sQ8_05tok/sQ8_0
DeepSeek R1 Distill Qwen 14B18tok/sQ8_05tok/sQ8_0
Qwen3.8 27B OBLITERATED14tok/smlx-4bit4tok/smlx-4bit
Devstral Small 250712tok/sQ8_04tok/sQ5_K_M
Magistral Small 250612tok/sQ8_04tok/sQ6_K
Magistral Small 250712tok/sQ8_04tok/sQ5_K_M
Magistral Small 250912tok/sQ8_04tok/sQ5_K_M
granite 4.2 30b11tok/sQ6_K4tok/sQ4_K_M
Muse Glimmer 30B9tok/sQ8_04tok/sQ5_K_M
Qwen 3.5 122B-A10Boffloadwon't run
Qwen3 Coder Nextoffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadwon't run
GLM 4.5 Airoffloadwon't run
Qwen3 Next 80B-A3Boffloadwon't run
Llama 3.3 70Boffloadwon't run
DeepSeek R1 Distill Llama 70Boffloadwon't run
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 B70 or the Mac mini M4 32GB?
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 B70 or the Mac mini M4 32GB?
At 8k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the Mac mini M4 32GB 58; only the Intel Arc Pro B70 holds Qwen 3.5 35B-A3B.
Which is faster for local AI, the Intel Arc Pro B70 or the Mac mini M4 32GB?
On the models both run fully in memory, the Intel Arc Pro B70 is about 3.9× as fast as the Mac mini M4 32GB (estimated, median across models).

More comparisons

More hardware for local AI