Rockchip RK3588 / Llama 3.1 8B

Can a Rockchip RK3588 run Llama 3.1 8B?

Technically yes, but slowly. Measured generation speed: 3.72 tok/s across 1 public run.

Rule of thumb: 10+ tok/s feels like a chat; 4–10 is usable for assistants and automations; under 4 is batch-job territory. Speeds depend on quantization, runtime, context length and cooling. Check each source.

BoardQuantRuntimeGen tok/sPrompt tok/sTTFT msSource
Rockchip RK3588
Indiedroid Nova (RK3588S, 16GB)
w8a8 (RKLLM conversion)RKLLM runtime 1.2.13.72––source ↗
repo 2026-01-12

Llama 3.1 8B on other boards

Buy a Rockchip RK3588

Ran Llama 3.1 8B on a Rockchip RK3588? Submit your numbers →

Get the monthly board report

New boards, new runtimes, new numbers. One email a month, no hype.