RTX 4070 Super 12GB vs RTX 4090D 24GB (China) for Local LLMs

The RTX 4090D 24GB (China) has about 100% more memory bandwidth (1,008 vs 504 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 100% faster on the RTX 4090D 24GB (China). The RTX 4090D 24GB (China) fits more of the catalog at Q4/8K — 33 vs 18 of 61 models (24 vs 12 GB VRAM). Largest model that fits the RTX 4090D 24GB (China) but not the RTX 4070 Super 12GB: GLM-4.7-Flash (31.2B). MSRP is $599 for the RTX 4070 Super 12GB 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 (40 shared models), the RTX 4090D 24GB (China) posts the higher tok/s in 33 rows vs 7. Methodology.

Spec comparison

	RTX 4070 Super 12GB	RTX 4090D 24GB (China)
Vendor	NVIDIA	NVIDIA
VRAM	12 GB	24 GB
Memory type	GDDR6X	GDDR6X
Memory bandwidth	504 GB/s	1,008 GB/s
TDP	220 W	425 W
MSRP (list)	$599	$1,599
Buy	Search 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 40 shared models

Model	Params (B)	RTX 4070 Super 12GB tok/s	RTX 4070 Super 12GB fits	RTX 4090D 24GB (China) tok/s	RTX 4090D 24GB (China) fits	Winner
Qwen3-0.6B	0.8	400.9 est.	Yes	801.8 est.	Yes	RTX 4090D 24GB (China)
gemma-3-1b-it	1.0	320.7 est.	Yes	641.5 est.	Yes	RTX 4090D 24GB (China)
Llama-3.2-1B-Instruct	1.2	267.3 est.	Yes	534.5 est.	Yes	RTX 4090D 24GB (China)
Qwen3-1.7B	2.0	160.4 est.	Yes	320.7 est.	Yes	RTX 4090D 24GB (China)
Qwen3-30B-A3B	30.5	130.2 est.	No	213.8 est.	Yes	RTX 4090D 24GB (China)
Qwen3-Coder-30B-A3B-Instruct	30.5	130.2 est.	No	213.8 est.	Yes	RTX 4090D 24GB (China)
GLM-4.7-Flash	31.2	125.0 est.	No	213.8 est.	Yes	RTX 4090D 24GB (China)
Qwen3-30B-A3B-Instruct-2507	30.5	130.2 est.	No	213.8 est.	Yes	RTX 4090D 24GB (China)
Llama-3.2-3B-Instruct	3.2	100.2 est.	Yes	200.5 est.	Yes	RTX 4090D 24GB (China)
gpt-oss-20b	21.5	199.6 est.	No	178.2 est.	Yes	RTX 4070 Super 12GB
Qwen3-Coder-Next	79.7	21.3 est.	No	170.4 est.	No	RTX 4090D 24GB (China)
Phi-4-mini-instruct	3.8	84.4 est.	Yes	168.8 est.	Yes	RTX 4090D 24GB (China)
Phi-4-mini-reasoning	3.8	84.4 est.	Yes	168.8 est.	Yes	RTX 4090D 24GB (China)
gemma-3-4b-it	4.3	82.2 est.	Yes	164.5 est.	Yes	RTX 4090D 24GB (China)
Qwen3-Next-80B-A3B-Instruct	81.3	20.5 est.	No	164.0 est.	No	RTX 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

	RTX 4070 Super 12GB	RTX 4090D 24GB (China)
Catalog models that fit	18 of 61	33 of 61
Largest exclusive fit	—	GLM-4.7-Flash (31.2B)

RTX 4070 Super 12GB details → · RTX 4090D 24GB (China) details →