← AI Hashrate

Best GPUs under $500 for local LLMs

15 catalog GPUs carry an MSRP under $500. They are ranked by decode speed (Q4 / 8K context) on Llama-3.1-8B-Instruct — a representative, widely used 8B-class model. MSRP is the launch list price, not a live price: street and Amazon prices move, so click through for the current number. Methodology.

#GPUMSRPVRAM (GB)Bandwidth (GB/s)Llama-3.1-8B-Instruct tok/sFits?Buy
1RTX 5060 Ti 16GB$4291644874.7 measuredYesSearch Amazon (affiliate)
2RX 7800 XT 16GB$4991662468.0 measuredYesSearch Amazon (affiliate)
3RTX 4060 Ti 16GB$4991628850.0 measuredYesSearch Amazon (affiliate)
4Arc A770 16GB$3491656044.5 est.YesSearch Amazon (affiliate)
5RTX 3060 12GB$3291236042.0 measuredYesMSI GeForce RTX 3060 Ventus 2X 12G OC (affiliate) · More listings
6Arc B580 12GB$2491245636.3 est.YesSearch Amazon (affiliate)
7RTX 3070 8GB$499844835.6 est.YesSearch Amazon (affiliate)
8RTX 5060 8GB$299844835.6 est.YesSearch Amazon (affiliate)
9RTX 5060 Ti 8GB$379844835.6 est.YesSearch Amazon (affiliate)
10RX 7700 XT 12GB$4491243234.4 est.YesSearch Amazon (affiliate)
11Arc B570 10GB$2191038030.2 est.YesSearch Amazon (affiliate)
12RTX 4060 Ti 8GB$399828822.9 est.YesSearch Amazon (affiliate)
13RX 7600 8GB$269828822.9 est.YesSearch Amazon (affiliate)
14RX 7600 XT 16GB$3291628822.9 est.YesSearch Amazon (affiliate)
15RTX 4060 8GB$299827221.6 est.YesSearch Amazon (affiliate)

tok/s figures are bandwidth-based estimates unless a row is tagged measured. The Fits? column links to the full Llama-3.1-8B-Instruct fit check on each card.

More lists: 7B-class models · 13B-class models · 30B-class models · 70B+ models · Under $1,000 · Under $2,000 · 8 GB VRAM · 12 GB VRAM · 16 GB VRAM · 24 GB VRAM · For Qwen3 · For DeepSeek · For Llama · For Gemma · For GLM · For coding · For reasoning · MoE picks