Hardware · Rockchip

What AI models can the Rockchip RK3576 (16 GB) run?

RAM: 16.0 GB
Usable for models: 13.6 GB
Memory bandwidth: 17 GB/s
Runs models on: CPU cores
Released: 2024

397 / 622 LLMs run Best: Gemma 4 12B 54 / 61 Image models run Best: FLUX 3 Image Klein 15 / 43 Video models run Best: LTX-Video 2B 65 / 70 Decision models run Best: OpenAI Privacy Filter

Models run on the CPU cores here, so the memory’s speed and how fast those cores can read it set the pace: small models (about 1-4B) are usable, bigger ones are slow. We plan around 13.6 GB of the 16.0 GB (what Linux leaves free for a model). 6 TOPS NPU (INT8), programmed with Rockchip’s RKNN and RKLLM toolkits. llama.cpp and Ollama use the CPU.

More about this hardware

Maker	Rockchip
Type	Single-board computer or edge AI board
Share of memory for models	85% (what Linux typically leaves free on a board this size)
Chip	Rockchip RK3576
AI accelerator	6 TOPS NPU (INT8), programmed with Rockchip’s RKNN and RKLLM toolkits. llama.cpp and Ollama use the CPU.
Boards with this chip	Seeed reComputer RK3576 · Radxa ROCK 4D · FriendlyElec NanoPi M5 · Luckfox Omni3576 · KICKPI K7
Memory type	LPDDR4X-4266 or LPDDR5-5500, 32-bit (17-22 GB/s)
Bandwidth figure	17.1 GB/s, the chip’s rated memory speed
Note	4× Cortex-A72 at 2.2 GHz plus 4× Cortex-A53, Mali-G52 GPU, up to 16 GB. Half the memory bus of the RK3588 (32-bit), so models generate at roughly half its speed. Speeds assume llama.cpp on the CPU at about 45% of the rated memory speed (estimate).
Specifications	www.rock-chips.com

Apps that run models well on the Rockchip RK3576 (16 GB)

Linux on ARM boards

For a typical model that fits in memory. Each model page picks the best app for that model on this hardware. All apps

Best LLMs for the Rockchip RK3576 (16 GB)

Quality	Model	Quant	Memory	Speed	Runs
80Great quality (A) · 80 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking.	Gemma 4 12BGoogle	`Q5_K_M`	10.9 GB	0.9 tok/s	Very slow
79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking.	K2 Horizon 7BInstitute of Foundation Models	`Q6_K`	10.3 GB	1.0 tok/s	Very slow
77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	MiMo V2.6 Distill Qwen 9BXiaomi	`Q6_K`	9.7 GB	1.0 tok/s	Very slow
77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	SenseNova Loop 9B PreviewSenseTime	`Q6_K`	10.1 GB	0.9 tok/s	Very slow
76Great quality (A) · 76 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	Ornith 1.5 9BDeepReinforce	`Q6_K`	9.9 GB	0.9 tok/s	Very slow
75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	Qwen3.5 9BAlibaba	`Q6_K`	9.9 GB	0.9 tok/s	Very slow
74Great quality (A) · 74 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	Ornith 1.0 9BDeepReinforce	`Q6_K`	9.9 GB	0.9 tok/s	Very slow
67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	Mellum2.1 Thinking (12B-A2.5B)JetBrains	`Q5_K_M`	10.4 GB	4.3 tok/s	Slow
73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	TMax 9BAi2	`Q6_K`	9.3 GB	1.0 tok/s	Very slow
64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking.	Gemma 4 26B-A4BGoogle	`Q2_K`	12.4 GB	5.6 tok/s	Slowtight fit
69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking.	UI-Mate 9BTencent	`Q6_K`	9.7 GB	1.0 tok/s	Very slow
69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking.	UI-Venus 2 9BAnt Group (inclusionAI)	`Q6_K`	9.7 GB	1.0 tok/s	Very slow

Best Image models for the Rockchip RK3576 (16 GB)

Quality	Model	Quant	Memory	Speed	Runs
90Excellent quality (A+) · 90 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	FLUX 3 Image KleinBlack Forest Labs	`Q8 GGUF`	12.8 GB	19.9 min	Very slowtight fit
76Great quality (A) · 76 of 100: how capable the model is at FP8, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking.	FLUX.2 [klein] 4BBlack Forest Labs	`FP8`	10.2 GB	2.7 min	Very slow
83Great quality (A) · 83 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	Z-Image TurboAlibaba (Tongyi)	`Q8 GGUF`	13.0 GB	8.7 min	Very slowtight fit
79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	ERNIE-Image TurboBaidu	`Q4 GGUF`	11.3 GB	11.6 min	Very slow
81Great quality (A) · 81 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking.	Z-ImageAlibaba (Tongyi)	`Q8 GGUF`	13.0 GB	59.3 min	Very slowtight fit
87Excellent quality (A+) · 87 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking.	ERNIE-ImageBaidu	`Q8 GGUF`	11.2 GB	80.5 min	Very slowencoder unloaded after use

Best Video models for the Rockchip RK3576 (16 GB)

Quality	Model	Quant	Memory	Speed	Runs
49Fair quality (C) · 49 of 100: how capable the model is at FP8, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking.	LTX-Video 2BLightricks	`FP8`	10.8 GB	68.2 min	Very slow
83Great quality (A) · 83 of 100: how capable the model is at FP8, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.	Bernini-R A14B (video generation + editing)ByteDance	`FP8`	10.7 GB	42.6 h	Very slowCPU offload
53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking.	Wan 2.1 T2V 1.3BAlibaba	`FP8`	12.8 GB	10.9 h	Very slowtight fit
44Fair quality (C) · 44 of 100: how capable the model is at FP8, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking.	CogVideoX 2BZhipu AI	`FP8`	11.0 GB	2.7 h	Very slow
76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking.	HY-OmniWeaving (8.3B, unified video)Tencent	`FP16`	10.7 GB	34.8 h	Very slowCPU offload

Best Decision models for the Rockchip RK3576 (16 GB)

Quality	Model	Quant	Memory	Speed	Runs
78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking.	OpenAI Privacy FilterOpenAI	`Q8_0`	2.4 GB	210 ms	Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	Qwen3 Reranker 4BAlibaba	`Q8_0`	6.3 GB	31.1 s	Very slow
90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking.	Qwen3 Reranker 8BAlibaba	`Q8_0`	10.7 GB	63.4 s	Very slow
90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	Skywork Reward V2 Llama 3.1 8BSkywork	`Q8_0`	10.4 GB	123.7 s	Very slow
86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	Jina Reranker v3.5Jina AI	`Q8_0`	2.6 GB	4.6 s	Very slow
88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	zerank-2ZeroEntropy	`Q8_0`	6.3 GB	31.1 s	Very slow
84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	Jina Reranker v3Jina AI	`Q8_0`	2.6 GB	4.6 s	Very slow
86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	Shieldstral 1.0 3BMistral AI	`Q8_0`	6.0 GB	29.8 s	Very slow

Best LLMs by task on this hardware Open in Model Finder

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.