← AI Hashrate

RTX 5080 16GB vs RX 9070 16GB for Local LLMs

The RTX 5080 16GB has about 50% more memory bandwidth (960 vs 640 GB/s). Batch-1 decode tracks bandwidth closely, so a model that fits both cards should decode roughly 50% faster on the RTX 5080 16GB. Both fit the same number of catalog models at Q4/8K (25 of 61; 16 vs 16 GB VRAM). MSRP is $999 for the RTX 5080 16GB vs $549 for the RX 9070 16GB. MSRP is a launch list price, not live retail — Amazon prices move; click through for the current price. In the Q4/8K head-to-head below (44 shared models), the RTX 5080 16GB posts the higher tok/s in 44 rows vs 0. Methodology.

Spec comparison

RTX 5080 16GBRX 9070 16GB
VendorNVIDIAAMD
VRAM16 GB16 GB
Memory typeGDDR7GDDR6
Memory bandwidth960 GB/s640 GB/s
TDP360 W220 W
MSRP (list)$999$549
BuyNVIDIA GeForce RTX 5080 Founders Edition (affiliate) · More listingsSearch 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 44 shared models

ModelParams (B)RTX 5080 16GB tok/sRTX 5080 16GB fitsRX 9070 16GB tok/sRX 9070 16GB fitsWinner
Qwen3-0.6B0.8763.6 est.Yes509.1 est.YesRTX 5080 16GB
gemma-3-1b-it1.0610.9 est.Yes407.3 est.YesRTX 5080 16GB
Llama-3.2-1B-Instruct1.2509.1 est.Yes339.4 est.YesRTX 5080 16GB
Qwen3-30B-A3B30.5440.9 est.No293.9 est.NoRTX 5080 16GB
Qwen3-Coder-30B-A3B-Instruct30.5440.9 est.No293.9 est.NoRTX 5080 16GB
Qwen3-30B-A3B-Instruct-250730.5440.9 est.No293.9 est.NoRTX 5080 16GB
GLM-4.7-Flash31.2423.2 est.No282.1 est.NoRTX 5080 16GB
Qwen3-1.7B2.0305.5 est.Yes203.6 est.YesRTX 5080 16GB
gpt-oss-20b21.5204.9 measuredYes113.1 est.YesRTX 5080 16GB
Llama-3.2-3B-Instruct3.2190.9 est.Yes127.3 est.YesRTX 5080 16GB
Mistral-7B-Instruct-v0.37.2163.6 measuredYes101.3 measuredYesRTX 5080 16GB
Phi-4-mini-instruct3.8160.8 est.Yes107.2 est.YesRTX 5080 16GB
Phi-4-mini-reasoning3.8160.8 est.Yes107.2 est.YesRTX 5080 16GB
gemma-3-4b-it4.3156.6 est.Yes104.4 est.YesRTX 5080 16GB
Qwen3-4B-Instruct-25074.0152.7 est.Yes101.8 est.YesRTX 5080 16GB

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 5080 16GBRX 9070 16GB
Catalog models that fit25 of 6125 of 61
Largest exclusive fit——

RTX 5080 16GB details → · RX 9070 16GB details →