H100 80GB SXM5 vs Tesla P4 8GB for Local LLMs

The H100 80GB SXM5 has about 1645% more memory bandwidth (3,350 vs 192 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 1645% faster on the H100 80GB SXM5. The H100 80GB SXM5 fits more of the catalog at Q4/8K — 50 vs 16 of 69 models (80 vs 8 GB VRAM). Largest model that fits the H100 80GB SXM5 but not the Tesla P4 8GB: Qwen3.5-122B-A10B (122.0B). MSRP is $30,000 for the H100 80GB SXM5 vs $80 for the Tesla P4 8GB. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. Not retail-shoppable: H100 80GB SXM5 (datacenter-class; cloud rental is the realistic way to use it). In the Q4/8K head-to-head below (39 shared models), the H100 80GB SXM5 posts the higher tok/s in 39 rows vs 0. Methodology.

Spec comparison

	H100 80GB SXM5	Tesla P4 8GB
Vendor	NVIDIA	NVIDIA
VRAM	80 GB	8 GB
Memory type	HBM3	GDDR5
Memory bandwidth	3,350 GB/s	192 GB/s
TDP	700 W	75 W
MSRP (list)	$30,000	$80
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 39 shared models

Model	Params (B)	H100 80GB SXM5 tok/s	H100 80GB SXM5 fits	Tesla P4 8GB tok/s	Tesla P4 8GB fits	Winner
MiniCPM5-1B	1.08	1973.9 est.	Yes	113.1 est.	Yes	H100 80GB SXM5
Qwen3.5-2B	2.0	1065.9 est.	Yes	61.1 est.	Yes	H100 80GB SXM5
DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx	2.3	926.9 est.	Yes	53.1 est.	Yes	H100 80GB SXM5
Qwen3.6-35B-A3B	36.0	710.6 est.	Yes	17.0 est.	No	H100 80GB SXM5
Qwen3.5-35B-A3B	34.7	710.6 est.	Yes	18.2 est.	No	H100 80GB SXM5
GLM-4.7-Flash	31.2	710.6 est.	Yes	21.0 est.	No	H100 80GB SXM5
Qwen3-30B-A3B-Instruct-2507	30.5	710.6 est.	Yes	21.9 est.	No	H100 80GB SXM5
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16	30.0	710.6 est.	Yes	24.1 est.	No	H100 80GB SXM5
Ornith-1.0-35B	35.0	710.6 est.	Yes	17.0 est.	No	H100 80GB SXM5
qwen35b-a3b-fable-sft-abliterated	34.7	710.6 est.	Yes	17.3 est.	No	H100 80GB SXM5
Llama-3.2-3B-Instruct	3.2	666.2 est.	Yes	38.2 est.	Yes	H100 80GB SXM5
gpt-oss-20b	21.5	592.2 est.	Yes	33.6 est.	No	H100 80GB SXM5
gemma-4-26B-A4B-it	25.2	561.0 est.	Yes	23.9 est.	No	H100 80GB SXM5
Qwen3-4B-Instruct-2507	4.0	533.0 est.	Yes	30.5 est.	Yes	H100 80GB SXM5
NVIDIA-Nemotron-3-Nano-4B-BF16	4.0	533.0 est.	Yes	30.5 est.	Yes	H100 80GB SXM5

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

	H100 80GB SXM5	Tesla P4 8GB
Catalog models that fit	50 of 69	16 of 69
Largest exclusive fit	Qwen3.5-122B-A10B (122.0B)	—

H100 80GB SXM5 details → · Tesla P4 8GB details →