← AI Hashrate

Arc B570 10GB vs RTX 4090D 24GB (China) for Local LLMs

The RTX 4090D 24GB (China) has about 165% more memory bandwidth (1,008 vs 380 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 165% faster on the RTX 4090D 24GB (China). The RTX 4090D 24GB (China) fits more of the catalog at Q4/8K — 33 vs 17 of 61 models (24 vs 10 GB VRAM). Largest model that fits the RTX 4090D 24GB (China) but not the Arc B570 10GB: GLM-4.7-Flash (31.2B). MSRP is $219 for the Arc B570 10GB vs $1,599 for the RTX 4090D 24GB (China). MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: RTX 4090D 24GB (China) (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (38 shared models), the RTX 4090D 24GB (China) posts the higher tok/s in 37 rows vs 1. Methodology.

Spec comparison

Arc B570 10GBRTX 4090D 24GB (China)
VendorIntelNVIDIA
VRAM10 GB24 GB
Memory typeGDDR6GDDR6X
Memory bandwidth380 GB/s1,008 GB/s
TDP150 W425 W
MSRP (list)$219$1,599
BuySearch Amazon (affiliate)

MSRP is the launch list price, not live retail. Buy links are Amazon affiliate links — the price after you click is the live price.

Q4 / 8K head-to-head — top 15 of 38 shared models

ModelParams (B)Arc B570 10GB tok/sArc B570 10GB fitsRTX 4090D 24GB (China) tok/sRTX 4090D 24GB (China) fitsWinner
Qwen3-0.6B0.8302.3 est.Yes801.8 est.YesRTX 4090D 24GB (China)
gemma-3-1b-it1.0241.8 est.Yes641.5 est.YesRTX 4090D 24GB (China)
Llama-3.2-1B-Instruct1.2201.5 est.Yes534.5 est.YesRTX 4090D 24GB (China)
Qwen3-1.7B2.0120.9 est.Yes320.7 est.YesRTX 4090D 24GB (China)
Qwen3-30B-A3B30.568.2 est.No213.8 est.YesRTX 4090D 24GB (China)
Qwen3-Coder-30B-A3B-Instruct30.568.2 est.No213.8 est.YesRTX 4090D 24GB (China)
GLM-4.7-Flash31.265.4 est.No213.8 est.YesRTX 4090D 24GB (China)
Qwen3-30B-A3B-Instruct-250730.568.2 est.No213.8 est.YesRTX 4090D 24GB (China)
Llama-3.2-3B-Instruct3.275.6 est.Yes200.5 est.YesRTX 4090D 24GB (China)
gpt-oss-20b21.5104.5 est.No178.2 est.YesRTX 4090D 24GB (China)
Phi-4-mini-instruct3.863.6 est.Yes168.8 est.YesRTX 4090D 24GB (China)
Phi-4-mini-reasoning3.863.6 est.Yes168.8 est.YesRTX 4090D 24GB (China)
gemma-3-4b-it4.362.0 est.Yes164.5 est.YesRTX 4090D 24GB (China)
Qwen3-4B-Instruct-25074.060.5 est.Yes160.4 est.YesRTX 4090D 24GB (China)
Qwen3-4B-Thinking-25074.060.5 est.Yes160.4 est.YesRTX 4090D 24GB (China)

est. = bandwidth-based estimate; measured = curated real benchmark that overrides the estimate. Fits = weights + KV(8K) + 1 GB overhead ≤ 95% of VRAM.

Fit coverage at Q4 / 8K

Arc B570 10GBRTX 4090D 24GB (China)
Catalog models that fit17 of 6133 of 61
Largest exclusive fitGLM-4.7-Flash (31.2B)

Arc B570 10GB details → · RTX 4090D 24GB (China) details →