readygpu

Intel Arc Pro B70 vs NVIDIA GeForce RTX 5070 Ti

For running models locally, priced in the UK. At 8k context the first runs 61 models fully in memory, the second 43.

In short

The Intel Arc Pro B70 has more memory for models: 32 GB against 16 GB. At 8k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the NVIDIA GeForce RTX 5070 Ti 43; only the Intel Arc Pro B70 holds Qwen 3.8 27B. On the models both run fully in memory, the NVIDIA GeForce RTX 5070 Ti is about 1.8× as fast as the Intel Arc Pro B70 (estimated, median across models).

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

Intel Arc Pro B70

Not priced here

Add to build
NVIDIA GeForce RTX 5070 Ti

£975

eBay

Add to build

Specifications

SpecIntel Arc Pro B70NVIDIA GeForce RTX 5070 Ti
GPUArc Pro B70GeForce RTX 5070 Ti
Memory32 GB GDDR616 GB GDDR7
Memory bus256-bit256-bit
Memory bandwidth608 GB/s896 GB/s
FP16 compute45.8 TFLOPS88 TFLOPS
SoftwareSYCL / oneAPICUDA
Board power230 W300 W
PCIe generationGen 5Gen 5
PCIe lanesx16x16

What each runs

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

ModelIntel Arc Pro B70NVIDIA GeForce RTX 5070 Ti
Qwen 3.8 27B10tok/sQ8_0offload
Qwen 3.6 27B10tok/sQ8_0offload
Gemma 4 31B11tok/sQ6_Koffload
Qwen 3.6 35B-A3B59tok/sQ6_Koffload
Qwen 3.5 27B10tok/sQ8_0offload
Gemma 4 26B-A4B37tok/sQ8_0offload
Qwen 3.5 35B-A3B63tok/sQ6_Koffload
GLM 4.7 Flash29tok/sQ8_0offload
Devstral Small 2 24B12tok/sQ8_035tok/sQ4_K_M
Nemotron 3.5 Lightning 30B-A3B56tok/sUD-Q5_K_Moffload
Kimi Linear 48B-A3B86tok/sQ4_K_Moffload
gpt-oss 20B64tok/sMXFP4127tok/sMXFP4
Qwen3 Coder 30B-A3B51tok/sQ6_Koffload
Qwen3 32B11tok/sQ6_Koffload
Gemma 4 12B23tok/sQ8_042tok/sQ8_0
Qwen 3.5 9B31tok/sQ8_056tok/sQ8_0
Mistral Small 3.2 24B12tok/sQ8_035tok/sQ4_K_M
DeepSeek R1 Distill Qwen 32B11tok/sQ6_Koffload
Gemma 3 27B10tok/sQ8_0offload
Ministral 3 14B19tok/sQ8_035tok/sQ8_0
Qwen3 14B18tok/sQ8_041tok/sQ6_K
DeepSeek R1 0528 Qwen3 8B31tok/sQ8_056tok/sQ8_0
Qwen3 8B31tok/sQ8_056tok/sQ8_0
Phi-4 14B18tok/sQ8_041tok/sQ6_K
Qwen 3.5 4B63tok/sQ8_0116tok/sQ8_0
Ministral 3 8B30tok/sQ8_055tok/sQ8_0
Granite 4.2 8B28tok/sQ8_052tok/sQ8_0
Gemma 4 E4B36tok/sQ8_066tok/sQ8_0
LFM2.5 8B-A1B114tok/sQ8_0227tok/sQ8_0
Llama 3.1 8B40tok/sQ6_K72tok/sQ6_K
Qwen2.5 0.5B481tok/sQ8_0879tok/sQ8_0
Qwen3.5 0.8B326tok/sQ8_0596tok/sQ8_0
Qwen3.5 2B196tok/sQ5_K_M357tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B143tok/sQ8_0261tok/sQ8_0
MiniCPM5 2B100tok/sQ8_0183tok/sQ8_0
LFM2.5 2.6B90tok/sQ8_0164tok/sQ8_0
LFM2.5 VL 3B90tok/sQ8_0164tok/sQ8_0
Spark X2.5 4B89tok/smlx-4bit163tok/smlx-4bit
granite vision 4.1 4b71tok/sQ8_0130tok/sQ8_0
Ministral 3 3B Instruct 251267tok/sQ8_0123tok/sQ8_0
granite 4.2 3b67tok/sQ8_0122tok/sQ8_0
FrogNano 4B 260962tok/sQ8_0112tok/sQ8_0
LFM2.5 24B A2B60tok/sQ8_0offload
Mellum2.1 12B A2.5B Thinking59tok/sQ8_0118tok/sQ8_0
Kimi VL A3B Thinking 250650tok/sQ8_0153tok/sQ4_K_M
Kimi VL A3B48tok/smlx-8bit120tok/smlx-6bit
DeepSeek R1 Distill Qwen 7B36tok/sQ8_065tok/sQ8_0
GLM 4 9B 041435tok/smlx-5bit65tok/smlx-5bit
DeepSeek R1 Distill Llama 8B32tok/sQ8_058tok/sQ8_0
Ministral 3 8B Reasoning 251230tok/sQ8_055tok/sQ8_0
GLM 4.6V Flash29tok/sQ8_054tok/sQ8_0
GLM 4.1V 9B Thinking25tok/smlx-8bit46tok/smlx-8bit
Ministral 3 14B Reasoning 251219tok/sQ8_035tok/sQ8_0
DeepSeek R1 Distill Qwen 14B18tok/sQ8_040tok/sQ6_K
Qwen3.8 27B OBLITERATED14tok/smlx-4bitoffload
Devstral Small 250712tok/sQ8_035tok/sQ4_K_M
Magistral Small 250612tok/sQ8_035tok/sQ4_K_M
Magistral Small 250712tok/sQ8_035tok/sQ4_K_M
Magistral Small 250912tok/sQ8_035tok/sQ4_K_M
granite 4.2 30b11tok/sQ6_Koffload
Muse Glimmer 30B9tok/sQ8_0offload
Qwen 3.5 122B-A10Boffloadoffload
Qwen3 Coder Nextoffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadoffload
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 NVIDIA GeForce RTX 5070 Ti?
Choose the Intel Arc Pro B70 to run bigger models. Choose the NVIDIA GeForce RTX 5070 Ti for speed.
Which can run bigger models, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5070 Ti?
At 8k context the Intel Arc Pro B70 runs 61 of the models we track fully in memory and the NVIDIA GeForce RTX 5070 Ti 43; only the Intel Arc Pro B70 holds Qwen 3.8 27B.
Which is faster for local AI, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5070 Ti?
On the models both run fully in memory, the NVIDIA GeForce RTX 5070 Ti is about 1.8× as fast as the Intel Arc Pro B70 (estimated, median across models).

More comparisons

More hardware for local AI