Which is better for local AI, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5090?

Choose the NVIDIA GeForce RTX 5090 for speed.

Intel Arc Pro B70 vs NVIDIA GeForce RTX 5090

For running models locally, priced in the US. At 64k context the first runs 58 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 NVIDIA GeForce RTX 5090 1,792 GB/s. At 64k context both run the same 58 models we track fully in memory. On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 3.7× as fast as the Intel Arc Pro B70 (estimated, median across models).

Updated October 26, 2026. Speeds are estimates; how they are worked out.

Intel Arc Pro B70

Not priced here

Add to build

NVIDIA GeForce RTX 5090

$4,921

eBay

Add to build

Specifications

Spec	Intel Arc Pro B70	NVIDIA GeForce RTX 5090
GPU	Arc Pro B70	GeForce RTX 5090
Memory	32 GB GDDR6	32 GB GDDR7
Memory bus	256-bit	512-bit
Memory bandwidth	608 GB/s	1,792 GB/s
FP16 compute	45.8 TFLOPS	209 TFLOPS
Software	SYCL / oneAPI	CUDA
Board power	230 W	575 W
PCIe generation	Gen 5	Gen 5
PCIe lanes	x16	x16

What each runs

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

Model	Intel Arc Pro B70	NVIDIA GeForce RTX 5090
Qwen 3.8 27B	11tok/sQ6_K	39tok/sQ6_K
Qwen 3.6 27B	11tok/sQ6_K	41tok/sQ6_K
Gemma 4 31B	10tok/sQ6_K	35tok/sQ6_K
Qwen 3.6 35B-A3B	43tok/sUD-Q6_K	167tok/sQ6_K
Qwen 3.5 27B	11tok/sQ6_K	41tok/sQ6_K
Gemma 4 26B-A4B	29tok/sQ8_0	117tok/sQ8_0
Qwen 3.5 35B-A3B	44tok/sQ6_K	175tok/sQ6_K
GLM 4.7 Flash	22tok/sQ6_K	87tok/sQ6_K
Devstral Small 2 24B	10tok/sQ6_K	37tok/sQ6_K
Nemotron 3.5 Lightning 30B-A3B	50tok/sUD-Q5_K_M	201tok/sUD-Q5_K_M
Kimi Linear 48B-A3B	70tok/sQ4_K_M	277tok/sQ4_K_M
gpt-oss 20B	41tok/sMXFP4	165tok/sMXFP4
Qwen3 Coder 30B-A3B	19tok/sQ6_K	75tok/sQ6_K
Gemma 4 12B	22tok/sQ8_0	79tok/sQ8_0
Qwen 3.5 9B	26tok/sQ8_0	95tok/sQ8_0
Gemma 3 27B	11tok/sQ6_K	40tok/sQ6_K
Mistral Small 3.2 24B	10tok/sQ6_K	37tok/sQ6_K
Ministral 3 14B	12tok/sQ8_0	44tok/sQ8_0
Qwen3 14B	12tok/sQ8_0	42tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B	17tok/sQ8_0	61tok/sQ8_0
Qwen3 8B	17tok/sQ8_0	61tok/sQ8_0
Phi-4 14B	11tok/sQ8_0	38tok/sQ8_0
Qwen 3.5 4B	46tok/sQ8_0	166tok/sQ8_0
Ministral 3 8B	17tok/sQ8_0	61tok/sQ8_0
Granite 4.2 8B	15tok/sQ8_0	55tok/sQ8_0
Gemma 4 E4B	30tok/sQ8_0	110tok/sQ8_0
LFM2.5 8B-A1B	39tok/sQ8_0	156tok/sQ8_0
Llama 3.1 8B	20tok/sQ6_K	73tok/sQ6_K
Qwen2.5 0.5B	228tok/sQ8_0	831tok/sQ8_0
Qwen3.5 0.8B	186tok/sQ8_0	679tok/sQ8_0
Qwen3.5 2B	135tok/sQ5_K_M	492tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B	81tok/sQ8_0	294tok/sQ8_0
MiniCPM5 2B	55tok/sQ8_0	202tok/sQ8_0
Spark X2.5 4B	55tok/smlx-4bit	202tok/smlx-4bit
Mellum2.1 12B A2.5B Thinking	46tok/sQ8_0	183tok/sQ8_0
FrogNano 4B 2609	45tok/sQ8_0	163tok/sQ8_0
LFM2.5 2.6B	44tok/sQ8_0	161tok/sQ8_0
LFM2.5 VL 3B	44tok/sQ8_0	161tok/sQ8_0
granite vision 4.1 4b	34tok/sQ8_0	124tok/sQ8_0
granite 4.2 3b	33tok/sQ8_0	120tok/sQ8_0
Kimi VL A3B Thinking 2506	33tok/sQ8_0	130tok/sQ8_0
Kimi VL A3B	32tok/smlx-8bit	126tok/smlx-8bit
Ministral 3 3B Instruct 2512	29tok/sQ8_0	105tok/sQ8_0
GLM 4 9B 0414	28tok/smlx-5bit	102tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B	26tok/sQ8_0	94tok/sQ8_0
GLM 4.6V Flash	24tok/sQ8_0	88tok/sQ8_0
LFM2.5 24B A2B	22tok/sQ8_0	89tok/sQ8_0
GLM 4.1V 9B Thinking	21tok/smlx-8bit	77tok/smlx-8bit
DeepSeek R1 Distill Llama 8B	18tok/sQ8_0	65tok/sQ8_0
Ministral 3 8B Reasoning 2512	17tok/sQ8_0	61tok/sQ8_0
Muse Glimmer 30B	13tok/sQ6_K	46tok/sQ6_K
Ministral 3 14B Reasoning 2512	12tok/sQ8_0	44tok/sQ8_0
Qwen3.8 27B OBLITERATED	12tok/smlx-4bit	44tok/smlx-4bit
Devstral Small 2507	11tok/sQ5_K_M	40tok/sQ5_K_M
Magistral Small 2507	11tok/sQ5_K_M	40tok/sQ5_K_M
Magistral Small 2509	11tok/sQ5_K_M	40tok/sQ5_K_M
DeepSeek R1 Distill Qwen 14B	11tok/sQ8_0	39tok/sQ8_0
Magistral Small 2506	10tok/sQ6_K	37tok/sQ6_K
Qwen 3.5 122B-A10B	offload	offload
Qwen3 Coder Next	offload	offload
Nemotron 3 Super 120B-A12B	offload	offload
Mistral Small 4 119B	offload	offload
GLM 4.5 Air	offload	offload
Qwen3 Next 80B-A3B	offload	offload
Qwen3 32B	offload	offload
DeepSeek R1 Distill Qwen 32B	offload	offload
granite 4.2 30b	offload	offload
DeepSeek V4 Flash	won't run	won't run
GLM 5.3 Flash	won't run	won't run
Qwen 3.8 Flash Next	won't run	won't run
MiniMax M2.7	won't run	won't run
gpt-oss 120B	won't run	won't run
Llama 3.3 70B	won't run	won't run
DeepSeek R1	won't run	won't run
DeepSeek R1 0528	won't run	won't run
DeepSeek R1 Distill Llama 70B	won't run	won't run
DeepSeek R1 Zero	won't run	won't run
DeepSeek V3	won't run	won't run
DeepSeek V3 0324	won't run	won't run
DeepSeek V3.1	won't run	won't run
DeepSeek V3.1 Terminus	won't run	won't run
DeepSeek V3.2	won't run	won't run
DeepSeek V4 Flash 0731	won't run	won't run
DeepSeek V4 Flash Vision Exp	won't run	won't run
Devstral 2 123B Instruct 2512	won't run	won't run
GLM 4.5	won't run	won't run
GLM 4.5V	won't run	won't run
GLM 4.6	won't run	won't run
GLM 4.6V	won't run	won't run
GLM 4.7	won't run	won't run
GLM 5	won't run	won't run
GLM 5.1	won't run	won't run
GLM 5.2	won't run	won't run
GLM 5.3	won't run	won't run
Kimi Dev 72B	won't run	won't run
Kimi K2	won't run	won't run
Kimi K2 Instruct 0905	won't run	won't run
Kimi K2 Thinking	won't run	won't run
Kimi K2.5	won't run	won't run
MiniMax M2	won't run	won't run
MiniMax M2.1	won't run	won't run
MiniMax M2.5	won't run	won't run
MiniMax M3	won't run	won't run
Mistral Medium 3.5 128B	won't run	won't run

Questions

Which is better for local AI, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5090?: Choose the NVIDIA GeForce RTX 5090 for speed.
Which can run bigger models, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5090?: At 64k context both run the same 58 models we track fully in memory.
Which is faster for local AI, the Intel Arc Pro B70 or the NVIDIA GeForce RTX 5090?: On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 3.7× as fast as the Intel Arc Pro B70 (estimated, median across models).

Intel Arc Pro B70 vs NVIDIA GeForce RTX 5090

In short

Specifications

What each runs

Questions

More comparisons

More hardware for local AI