What are the best models to run on AMD RX 7600 XT (16GB)?

Based on model rankings and 16 GB usable VRAM: Top Overall: Qwen 3.8 27B (INT4) ~14 tok/s. Top Coding: Qwen3.5-27B (INT4) ~14 tok/s. Optimal Daily Driver: Phi-4 Reasoning Plus (INT4) ~20 tok/s. Maximum Model Size: ERNIE-4.5-VL-28B-A3B (INT4) ~73 tok/s.

How are VRAM, throughput, and latency calculated?

Calculations are determined using apxml.com modeling, combining LLM architecture analysis with GPU hardware physics. VRAM and Context Scaling: Memory requirements are determined from model architecture, including parameter count, layer configuration, and attention structure. The KV cache footprint scales deterministically based on context length, batch size, and cache precision. Throughput (TPS): Token generation is calculated in the memory-bandwidth-bound decode phase, measuring the time required to stream model weights and KV cache relative to hardware memory bandwidth. Latency (TTFT): Time to first token is calculated from the compute operations required during prompt prefill relative to hardware tensor compute capacity.

How accurate are these estimates?

Calculations are calibrated against empirical ground-truth benchmark runs across diverse architectures and hardware. Because the underlying mechanics are benchmark-calibrated, estimates remain accurate across different configurations without needing to run physical tests for every scenario. VRAM Accuracy: Memory estimates are deterministic and closely match actual allocation. Minor differences typically stay within 3% to 5% due to CUDA runtime overhead or memory allocator fragmentation. Speed and Latency Accuracy: Throughput and TTFT typically vary within 5% to 10% depending on the specific inference runtime, kernel implementations, and system power limits.

Best Models to Run on AMD RX 7600 XT (16GB) (October 2026)

Rank	Model	Parameters	Precision	TPS (Approx.)	TTFT
#33	Qwen 3.8 27B	27B	INT4	14 tok/s	~7.7s
#83	Qwen3.5-27B	27B	INT4	14 tok/s	~7.7s
#100	Phi-4 Reasoning Plus	14B	INT4	20 tok/s	~3.4s
#101	MiMo V2 Flash	15B (309B active)	INT4	1 tok/s	~77.6s
#108	Gemma 4 26B A4B	25.2B (3.8B active)	INT4	68 tok/s	~1.3s
#109	Gemma 4 12B	11.95B	INT4 Q8	23 tok/s 14 tok/s	~3.0s ~3.5s
#113	Qwen3.5-9B	9B	INT4 Q8	42 tok/s 25 tok/s	~2.2s ~2.2s
#118	K2 Horizon 7B	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
#128	NVIDIA Nemotron 3 Nano 30B-A3B	3.5B (30B active)	INT4 Q8 FP16	13 tok/s 8 tok/s 4 tok/s	~7.0s ~7.1s ~7.2s
#132	Qwen3.5-4B	4B	INT4 Q8 FP16	80 tok/s 50 tok/s 28 tok/s	~1.0s ~1.0s ~1.0s
#136	GPT-OSS 20B	21B (3.6B active)	INT4	29 tok/s	~1.3s
#146	Granite 4.2 8B	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#150	Gemma 4 E4B	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#158	GLM-4V	9B	INT4 Q8	29 tok/s 19 tok/s	~2.2s ~2.4s
#159	Gemma 3 27B	27B	INT4	14 tok/s	~7.7s
#160	Phi-4	14B	INT4 Q8	28 tok/s 15 tok/s	~3.4s ~4.1s
#169	OLMo 3 7B Think	7B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
#172	DeepSeek-R1 14B	14B	INT4 Q8	29 tok/s 15 tok/s	~3.4s ~4.1s
#173	Ministral 3 14B	14B	INT4	20 tok/s	~3.4s
#177	Gemma 3 12B	12B	INT4 Q8	32 tok/s 18 tok/s	~3.0s ~3.2s
#181	Mistral-Small-2501	24B	INT4	16 tok/s	~6.3s
#182	Ministral 3 8B	8B	INT4 Q8	33 tok/s 22 tok/s	~2.0s ~2.0s
#183	Qwen3.5-2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~545 ms ~549 ms ~557 ms
#184	MiniCPM5-2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~545 ms ~549 ms ~557 ms
#185	Qwen2.5-14B	14B	INT4 Q8	28 tok/s 15 tok/s	~3.4s ~4.1s
#186	Gemma 2 27B	27B	INT4	14 tok/s	~7.7s
#188	Gemma 4 E2B	5.1B	INT4 Q8 FP16	67 tok/s 41 tok/s 22 tok/s	~1.3s ~1.3s ~1.3s
#189	Qwen3-4B	4B	INT4 Q8 FP16	80 tok/s 50 tok/s 28 tok/s	~1.0s ~1.0s ~1.0s
#190	DeepSeek-R1 8B	8B	INT4 Q8	47 tok/s 28 tok/s	~2.0s ~2.0s
#193	Ministral 3 3B	3B	INT4 Q8 FP16	63 tok/s 48 tok/s 30 tok/s	~789 ms ~794 ms ~806 ms
#195	Qwen3-14B	14B	INT4 Q8	28 tok/s 15 tok/s	~3.4s ~4.1s
#197	Gemma 2 9B	9B	INT4 Q8	42 tok/s 25 tok/s	~2.2s ~2.2s
#198	Qwen2.5-7B	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
#199	Qwen3-8B	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#200	Gemma 3 4B	4B	INT4 Q8 FP16	80 tok/s 50 tok/s 28 tok/s	~1.0s ~1.0s ~1.0s
#201	DeepSeek-R1 7B	7B	INT4 Q8 FP16	53 tok/s 31 tok/s 16 tok/s	~1.7s ~1.8s ~1.9s
#202	Llama 3.1 8B	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#204	Qwen3-1.7B	1.7B	INT4 Q8 FP16	140 tok/s 98 tok/s 58 tok/s	~467 ms ~470 ms ~477 ms
#206	Gemma 3 270M	0.27B	INT4 Q8 FP16	256 tok/s 229 tok/s 183 tok/s	~78 ms ~79 ms ~80 ms
#208	Llama 3 8B	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#209	Qwen3-0.6B	0.6B	INT4 Q8 FP16	236 tok/s 187 tok/s 128 tok/s	~167 ms ~169 ms ~171 ms
#210	Phi-4-Mini	3.8B	INT4 Q8 FP16	83 tok/s 53 tok/s 29 tok/s	~975 ms ~982 ms ~998 ms
#212	Ministral-8B-2410	8B	INT4 Q8	46 tok/s 27 tok/s	~2.0s ~2.0s
#213	DeepSeek-R1 1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~414 ms ~417 ms ~423 ms
#214	Llama 3.2 3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 36 tok/s	~783 ms ~789 ms ~801 ms
#215	Qwen3.5-0.8B	0.8B	INT4 Q8 FP16	209 tok/s 160 tok/s 105 tok/s	~225 ms ~226 ms ~229 ms
#216	Phi-3-mini	3.8B	INT4 Q8 FP16	83 tok/s 53 tok/s 29 tok/s	~975 ms ~982 ms ~998 ms
#217	Gemma 3n E2B IT	6B (2B active)	INT4 Q8 FP16	63 tok/s 53 tok/s 35 tok/s	~590 ms ~593 ms ~652 ms
#218	Mistral-7B-Instruct-v0.1	7.3B	INT4 Q8 FP16	50 tok/s 30 tok/s 14 tok/s	~1.8s ~1.8s ~2.2s
#219	Llama 3.2 1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~284 ms ~286 ms ~290 ms
#220	Gemma 2 2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~545 ms ~549 ms ~557 ms
#221	Phi-3-medium	14B	INT4 Q8	28 tok/s 15 tok/s	~3.4s ~4.1s
#222	OLMo 3 7B Instruct	7B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
#224	Qwen2-7B	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
#225	Phi-3-small	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
#226	Gemma 3 1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~284 ms ~286 ms ~290 ms
#227	Mistral-7B-Instruct-v0.2	7.3B	INT4 Q8 FP16	50 tok/s 30 tok/s 14 tok/s	~1.8s ~1.8s ~2.2s
#229	Gemma 1 7B	7B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
#230	Gemma 1 2B	2B	INT4 Q8 FP16	137 tok/s 91 tok/s 52 tok/s	~545 ms ~548 ms ~557 ms
#231	ChatGLM3-6B	6B	INT4 Q8 FP16	40 tok/s 28 tok/s 16 tok/s	~1.5s ~1.5s ~1.7s
#232	ChatGLM2-6B	6B	INT4 Q8 FP16	40 tok/s 28 tok/s 16 tok/s	~1.5s ~1.5s ~1.7s
#233	ChatGLM-6B	6B	INT4 Q8 FP16	40 tok/s 28 tok/s 16 tok/s	~1.5s ~1.5s ~1.7s
Unranked Models (Sorted by Speed)
-	Qwen2-0.5B	0.5B	INT4 Q8 FP16	252 tok/s 204 tok/s 144 tok/s	~140 ms ~141 ms ~143 ms
-	Qwen2.5-0.5B	0.5B	INT4 Q8 FP16	252 tok/s 204 tok/s 144 tok/s	~140 ms ~141 ms ~143 ms
-	ERNIE-4.5-0.3B	0.3B	INT4 Q8 FP16	245 tok/s 218 tok/s 172 tok/s	~87 ms ~87 ms ~88 ms
-	ERNIE-4.5-0.3B-Base	0.3B	INT4 Q8 FP16	245 tok/s 218 tok/s 172 tok/s	~87 ms ~87 ms ~88 ms
-	Falcon-1B	1B	INT4 Q8 FP16	199 tok/s 145 tok/s 91 tok/s	~284 ms ~286 ms ~290 ms
-	Falcon3-1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~284 ms ~286 ms ~290 ms
-	Qwen2-1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~414 ms ~417 ms ~423 ms
-	Qwen2.5-1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~414 ms ~417 ms ~423 ms
-	Spark X2.5 1.7B	1.7B	INT4 Q8 FP16	140 tok/s 98 tok/s 58 tok/s	~467 ms ~470 ms ~477 ms
-	Hy-MT2 1.8B	1.8B	INT4 Q8 FP16	136 tok/s 94 tok/s 55 tok/s	~492 ms ~495 ms ~502 ms
-	Falcon-3B	3B	INT4 Q8 FP16	105 tok/s 66 tok/s 37 tok/s	~783 ms ~788 ms ~801 ms
-	CroissantLLM Base	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~365 ms ~367 ms ~372 ms
-	Phi-1	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~365 ms ~367 ms ~372 ms
-	Phi-1.5	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~365 ms ~367 ms ~372 ms
-	DeepSeek-R1 3B	3B	INT4 Q8 FP16	100 tok/s 64 tok/s 36 tok/s	~783 ms ~789 ms ~801 ms
-	Falcon3-3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 36 tok/s	~783 ms ~789 ms ~801 ms
-	Qwen2.5-3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 36 tok/s	~783 ms ~789 ms ~801 ms
-	ERNIE-4.5-21B-A3B	21B (3B active)	INT4	80 tok/s	~1.0s
-	ERNIE-4.5-21B-A3B-Base	21B (3B active)	INT4	80 tok/s	~1.0s
-	Spark X2.5 4B	4B	INT4 Q8 FP16	80 tok/s 50 tok/s 28 tok/s	~1.0s ~1.0s ~1.0s
-	ERNIE-4.5-VL-28B-A3B	28B (3B active)	INT4	73 tok/s	~1.2s
-	ERNIE-4.5-VL-28B-A3B-Base	28B (3B active)	INT4	73 tok/s	~1.2s
-	Phi-2	2.7B	INT4 Q8 FP16	66 tok/s 51 tok/s 33 tok/s	~718 ms ~722 ms ~733 ms
-	MaLLaM-3B	3B	INT4 Q8 FP16	63 tok/s 48 tok/s 30 tok/s	~789 ms ~794 ms ~806 ms
-	SmolLM3 3B	3B	INT4 Q8 FP16	63 tok/s 48 tok/s 30 tok/s	~789 ms ~794 ms ~806 ms
-	Falcon-7B	7B	INT4 Q8 FP16	54 tok/s 32 tok/s 16 tok/s	~1.7s ~1.8s ~1.9s
-	Falcon3-7B	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
-	Hy-MT2 7B	7B	INT4 Q8 FP16	51 tok/s 31 tok/s 15 tok/s	~1.7s ~1.8s ~2.1s
-	Mistral-7B-v0.1	7.3B	INT4 Q8 FP16	50 tok/s 30 tok/s 14 tok/s	~1.8s ~1.8s ~2.2s
-	LensVLM 9B	9B	INT4 Q8	42 tok/s 25 tok/s	~2.2s ~2.2s
-	MiMo V2.6 Distill Qwen 9B	9B	INT4 Q8	42 tok/s 25 tok/s	~2.2s ~2.2s
-	ChatGLM3-6B-32K	6B	INT4 Q8 FP16	40 tok/s 28 tok/s 16 tok/s	~1.5s ~1.5s ~1.7s
-	Yi-6B	6B	INT4 Q8 FP16	40 tok/s 28 tok/s 16 tok/s	~1.5s ~1.5s ~1.7s
-	Falcon3-10B	10B	INT4 Q8	38 tok/s 22 tok/s	~2.5s ~2.5s
-	Falcon2-11B	11B	INT4 Q8	36 tok/s 21 tok/s	~2.7s ~2.7s
-	Hunyuan Lite	7B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
-	Kimi-VL-A3B-Instruct	16B (3B active)	INT4	36 tok/s	~1.0s
-	OLMo 3 7B Base	7B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
-	SEA-LION-7B	7.1B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
-	SEA-LION-7B-Instruct	7.1B	INT4 Q8	36 tok/s 25 tok/s	~1.8s ~1.8s
-	Sahabat-AI-Llama3-8B-Instruct	8B	INT4 Q8	33 tok/s 22 tok/s	~2.0s ~2.0s
-	Typhoon-2-8B	8B	INT4 Q8	33 tok/s 22 tok/s	~2.0s ~2.0s
-	GLM-4-9B	9B	INT4 Q8	29 tok/s 19 tok/s	~2.2s ~2.4s
-	GLM-4-9B-Chat	9B	INT4 Q8	29 tok/s 19 tok/s	~2.2s ~2.4s
-	GLM-4-9B-Chat-1M	9B	INT4 Q8	29 tok/s 19 tok/s	~2.2s ~2.4s
-	Sahabat-AI-Gemma2-9B	9.2B	INT4 Q8	29 tok/s 19 tok/s	~2.3s ~2.5s
-	Sahabat-AI-Gemma2-9B-Instruct	9.2B	INT4 Q8	29 tok/s 19 tok/s	~2.3s ~2.5s
-	Yi-9B	9B	INT4 Q8	29 tok/s 19 tok/s	~2.2s ~2.4s
-	GreenMind-14B-R1	14B	INT4	20 tok/s	~3.4s

AMD RX 7600 XT (16GB)

VRAM Scaling by Context Length

Best Models to Run on AMD RX 7600 XT (16GB)

Inference Capacity

Fine-Tuning Capacity

Workload Recommendations

Frequently Asked Questions