← AI Hashrate 中文

MacBook Neo A18 Pro 8GB vs RTX 4090D 24GB (China) for Local LLMs

The RTX 4090D 24GB (China) has about 1580% more memory bandwidth (1,008 vs 60 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 1580% faster on the RTX 4090D 24GB (China). The RTX 4090D 24GB (China) fits more of the catalog at Q4/8K — 36 vs 11 of 67 models (24 vs 8 GB VRAM). Largest model that fits the RTX 4090D 24GB (China) but not the MacBook Neo A18 Pro 8GB: Qwen3.6-35B-A3B (36.0B). MSRP is $0 for the MacBook Neo A18 Pro 8GB 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 (36 shared models), the RTX 4090D 24GB (China) posts the higher tok/s in 36 rows vs 0. Methodology.

Spec comparison

MacBook Neo A18 Pro 8GBRTX 4090D 24GB (China)
VendorAppleNVIDIA
VRAM8 GB24 GB
Memory typeLPDDR5XGDDR6X
Memory bandwidth60 GB/s1,008 GB/s
TDP20 W425 W
MSRP (list)$0$1,599
BuySearch Amazon—

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 36 shared models

ModelParams (B)MacBook Neo A18 Pro 8GB tok/sMacBook Neo A18 Pro 8GB fitsRTX 4090D 24GB (China) tok/sRTX 4090D 24GB (China) fitsWinner
MiniCPM5-1B1.0870.0 est.Yes797.2 est.YesRTX 4090D 24GB (China)
Qwen3.5-2B2.037.8 est.Yes430.5 est.YesRTX 4090D 24GB (China)
Qwen3.6-35B-A3B36.02.6 est.No287.0 est.YesRTX 4090D 24GB (China)
Qwen3.5-35B-A3B34.72.8 est.No287.0 est.YesRTX 4090D 24GB (China)
GLM-4.7-Flash31.22.5 est.No287.0 est.YesRTX 4090D 24GB (China)
Qwen3-30B-A3B-Instruct-250730.53.3 est.No287.0 est.YesRTX 4090D 24GB (China)
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF1630.03.5 est.No287.0 est.YesRTX 4090D 24GB (China)
Ornith-1.0-35B35.02.7 est.No287.0 est.YesRTX 4090D 24GB (China)
Llama-3.2-3B-Instruct3.223.6 est.Yes269.0 est.YesRTX 4090D 24GB (China)
gpt-oss-20b21.55.7 est.No239.1 est.YesRTX 4090D 24GB (China)
gemma-4-26B-A4B-it25.23.9 est.No226.6 est.YesRTX 4090D 24GB (China)
Nanbeige4.2-3B4.019.2 est.Yes219.1 est.YesRTX 4090D 24GB (China)
Qwen3-4B-Instruct-25074.018.9 est.Yes215.2 est.YesRTX 4090D 24GB (China)
NVIDIA-Nemotron-3-Nano-4B-BF164.018.9 est.Yes215.2 est.YesRTX 4090D 24GB (China)
Qwen3-4B-Thinking-25074.018.9 est.Yes215.2 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

MacBook Neo A18 Pro 8GBRTX 4090D 24GB (China)
Catalog models that fit11 of 6736 of 67
Largest exclusive fit—Qwen3.6-35B-A3B (36.0B)

MacBook Neo A18 Pro 8GB details → · RTX 4090D 24GB (China) details →