What are the best models to run on AMD RX 9060 XT (16GB)?

Based on model rankings and 16 GB usable VRAM: Top Overall: Qwen 3.8 27B (INT4) ~15 tok/s. Top Coding: Gemma 4 26B A4B (INT4) ~74 tok/s. Optimal Daily Driver: Phi-4 Reasoning Plus (INT4) ~22 tok/s. Maximum Model Size: ERNIE-4.5-VL-28B-A3B (INT4) ~79 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 9060 XT (16GB) (October 2026)

Rank	Model	Parameters	Precision	TPS (Approx.)	TTFT
#32	Qwen 3.8 27B	27B	INT4	15 tok/s	~2.9s
#82	Qwen3.5-27B	27B	INT4	15 tok/s	~2.9s
#96	Phi-4 Reasoning Plus	14B	INT4	22 tok/s	~1.3s
#97	MiMo V2 Flash	15B (309B active)	INT4	1 tok/s	~29.6s
#110	Gemma 4 12B	11.95B	INT4 Q8	26 tok/s 15 tok/s	~1.1s ~1.4s
#116	Gemma 4 26B A4B	25.2B (3.8B active)	INT4	74 tok/s	~490 ms
#118	K2 Horizon 7B	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
#132	Qwen3.5-4B	4B	INT4 Q8 FP16	87 tok/s 55 tok/s 30 tok/s	~395 ms ~402 ms ~416 ms
#133	Qwen3.5-9B	9B	INT4 Q8	46 tok/s 27 tok/s	~854 ms ~869 ms
#136	GPT-OSS 20B	21B (3.6B active)	INT4	32 tok/s	~512 ms
#141	NVIDIA Nemotron 3 Nano 30B-A3B	3.5B (30B active)	INT4 Q8 FP16	15 tok/s 8 tok/s 4 tok/s	~2.7s ~2.7s ~2.8s
#149	Granite 4.2 8B	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#155	Gemma 4 E4B	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#159	Gemma 3 27B	27B	INT4	15 tok/s	~2.9s
#161	Phi-4	14B	INT4 Q8	31 tok/s 16 tok/s	~1.3s ~1.6s
#163	GLM-4V	9B	INT4 Q8	32 tok/s 21 tok/s	~863 ms ~951 ms
#173	Ministral 3 14B	14B	INT4	22 tok/s	~1.3s
#176	DeepSeek-R1 14B	14B	INT4 Q8	32 tok/s 16 tok/s	~1.3s ~1.6s
#177	Gemma 3 12B	12B	INT4 Q8	35 tok/s 20 tok/s	~1.1s ~1.2s
#182	Mistral-Small-2501	24B	INT4	18 tok/s	~2.4s
#183	Ministral 3 8B	8B	INT4 Q8	36 tok/s 25 tok/s	~769 ms ~781 ms
#184	Qwen3.5-2B	2B	INT4 Q8 FP16	136 tok/s 94 tok/s 55 tok/s	~213 ms ~216 ms ~223 ms
#185	Qwen2.5-14B	14B	INT4 Q8	31 tok/s 16 tok/s	~1.3s ~1.6s
#186	MiniCPM5-2B	2B	INT4 Q8 FP16	136 tok/s 94 tok/s 55 tok/s	~213 ms ~216 ms ~223 ms
#187	Gemma 2 27B	27B	INT4	15 tok/s	~2.9s
#188	OLMo 3 7B Think	7B	INT4 Q8	40 tok/s 28 tok/s	~677 ms ~688 ms
#190	Gemma 4 E2B	5.1B	INT4 Q8 FP16	72 tok/s 45 tok/s 24 tok/s	~496 ms ~504 ms ~523 ms
#192	Ministral 3 3B	3B	INT4 Q8 FP16	69 tok/s 52 tok/s 33 tok/s	~309 ms ~313 ms ~324 ms
#194	DeepSeek-R1 8B	8B	INT4 Q8	52 tok/s 31 tok/s	~760 ms ~774 ms
#195	Gemma 2 9B	9B	INT4 Q8	46 tok/s 27 tok/s	~854 ms ~869 ms
#197	Gemma 3 4B	4B	INT4 Q8 FP16	87 tok/s 55 tok/s 30 tok/s	~395 ms ~402 ms ~416 ms
#198	Qwen2.5-7B	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
#199	DeepSeek-R1 7B	7B	INT4 Q8 FP16	58 tok/s 34 tok/s 17 tok/s	~669 ms ~681 ms ~765 ms
#199	Qwen3-14B	14B	INT4 Q8	31 tok/s 16 tok/s	~1.3s ~1.6s
#202	Qwen3-8B	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#203	Llama 3.1 8B	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#204	Qwen3-4B	4B	INT4 Q8 FP16	87 tok/s 55 tok/s 30 tok/s	~395 ms ~402 ms ~416 ms
#209	Llama 3 8B	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#210	Phi-4-Mini	3.8B	INT4 Q8 FP16	90 tok/s 57 tok/s 32 tok/s	~377 ms ~383 ms ~397 ms
#211	Gemma 3 270M	0.27B	INT4 Q8 FP16	264 tok/s 238 tok/s 193 tok/s	~35 ms ~35 ms ~36 ms
#212	Qwen3-1.7B	1.7B	INT4 Q8 FP16	150 tok/s 105 tok/s 64 tok/s	~183 ms ~186 ms ~192 ms
#213	Qwen3-0.6B	0.6B	INT4 Q8 FP16	245 tok/s 197 tok/s 137 tok/s	~69 ms ~70 ms ~72 ms
#214	Ministral-8B-2410	8B	INT4 Q8	50 tok/s 30 tok/s	~761 ms ~774 ms
#216	DeepSeek-R1 1.5B	1.5B	INT4 Q8 FP16	161 tok/s 115 tok/s 70 tok/s	~162 ms ~165 ms ~170 ms
#217	Llama 3.2 3B	3B	INT4 Q8 FP16	106 tok/s 69 tok/s 39 tok/s	~303 ms ~308 ms ~319 ms
#218	Phi-3-mini	3.8B	INT4 Q8 FP16	90 tok/s 57 tok/s 32 tok/s	~377 ms ~383 ms ~397 ms
#219	Qwen3.5-0.8B	0.8B	INT4 Q8 FP16	218 tok/s 169 tok/s 113 tok/s	~90 ms ~92 ms ~95 ms
#220	Gemma 3n E2B IT	6B (2B active)	INT4 Q8 FP16	68 tok/s 58 tok/s 38 tok/s	~234 ms ~236 ms ~263 ms
#221	Gemma 2 2B	2B	INT4 Q8 FP16	136 tok/s 94 tok/s 55 tok/s	~213 ms ~216 ms ~223 ms
#222	Phi-3-medium	14B	INT4 Q8	31 tok/s 16 tok/s	~1.3s ~1.6s
#223	Mistral-7B-Instruct-v0.1	7.3B	INT4 Q8 FP16	54 tok/s 33 tok/s 16 tok/s	~696 ms ~708 ms ~860 ms
#224	Llama 3.2 1B	1B	INT4 Q8 FP16	196 tok/s 148 tok/s 96 tok/s	~113 ms ~115 ms ~118 ms
#226	OLMo 3 7B Instruct	7B	INT4 Q8	40 tok/s 28 tok/s	~677 ms ~688 ms
#227	Qwen2-7B	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
#228	Phi-3-small	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
#229	Gemma 3 1B	1B	INT4 Q8 FP16	196 tok/s 148 tok/s 96 tok/s	~113 ms ~115 ms ~118 ms
#230	Mistral-7B-Instruct-v0.2	7.3B	INT4 Q8 FP16	54 tok/s 33 tok/s 16 tok/s	~696 ms ~708 ms ~860 ms
#232	Gemma 1 7B	7B	INT4 Q8	40 tok/s 28 tok/s	~677 ms ~688 ms
#233	Gemma 1 2B	2B	INT4 Q8 FP16	146 tok/s 98 tok/s 57 tok/s	~212 ms ~215 ms ~223 ms
#234	ChatGLM3-6B	6B	INT4 Q8 FP16	44 tok/s 31 tok/s 18 tok/s	~585 ms ~594 ms ~667 ms
#235	ChatGLM2-6B	6B	INT4 Q8 FP16	44 tok/s 31 tok/s 18 tok/s	~585 ms ~594 ms ~667 ms
#236	ChatGLM-6B	6B	INT4 Q8 FP16	44 tok/s 31 tok/s 18 tok/s	~585 ms ~594 ms ~667 ms
Unranked Models (Sorted by Speed)
-	Qwen2-0.5B	0.5B	INT4 Q8 FP16	260 tok/s 214 tok/s 154 tok/s	~58 ms ~59 ms ~61 ms
-	Qwen2.5-0.5B	0.5B	INT4 Q8 FP16	260 tok/s 214 tok/s 154 tok/s	~58 ms ~59 ms ~61 ms
-	ERNIE-4.5-0.3B	0.3B	INT4 Q8 FP16	254 tok/s 228 tok/s 182 tok/s	~38 ms ~39 ms ~40 ms
-	ERNIE-4.5-0.3B-Base	0.3B	INT4 Q8 FP16	254 tok/s 228 tok/s 182 tok/s	~38 ms ~39 ms ~40 ms
-	Liquid D1-Omni 600M	0.6B	INT4 Q8 FP16	245 tok/s 197 tok/s 137 tok/s	~69 ms ~70 ms ~72 ms
-	Falcon-1B	1B	INT4 Q8 FP16	209 tok/s 155 tok/s 99 tok/s	~113 ms ~115 ms ~118 ms
-	Falcon3-1B	1B	INT4 Q8 FP16	196 tok/s 148 tok/s 96 tok/s	~113 ms ~115 ms ~118 ms
-	Qwen2-1.5B	1.5B	INT4 Q8 FP16	161 tok/s 115 tok/s 70 tok/s	~162 ms ~165 ms ~170 ms
-	Qwen2.5-1.5B	1.5B	INT4 Q8 FP16	161 tok/s 115 tok/s 70 tok/s	~162 ms ~165 ms ~170 ms
-	Spark X2.5 1.7B	1.7B	INT4 Q8 FP16	150 tok/s 105 tok/s 64 tok/s	~183 ms ~186 ms ~192 ms
-	Hy-MT2 1.8B	1.8B	INT4 Q8 FP16	145 tok/s 101 tok/s 61 tok/s	~192 ms ~195 ms ~202 ms
-	Falcon-3B	3B	INT4 Q8 FP16	113 tok/s 72 tok/s 40 tok/s	~303 ms ~308 ms ~319 ms
-	CroissantLLM Base	1.3B	INT4 Q8 FP16	111 tok/s 92 tok/s 64 tok/s	~146 ms ~147 ms ~152 ms
-	Phi-1	1.3B	INT4 Q8 FP16	111 tok/s 92 tok/s 64 tok/s	~146 ms ~147 ms ~152 ms
-	Phi-1.5	1.3B	INT4 Q8 FP16	111 tok/s 92 tok/s 64 tok/s	~146 ms ~147 ms ~152 ms
-	DeepSeek-R1 3B	3B	INT4 Q8 FP16	108 tok/s 70 tok/s 40 tok/s	~303 ms ~308 ms ~319 ms
-	Mellum 2.1 12B A2.5B Thinking	12B (2.5B active)	INT4 Q8	108 tok/s 71 tok/s	~296 ms ~325 ms
-	Falcon3-3B	3B	INT4 Q8 FP16	106 tok/s 69 tok/s 39 tok/s	~303 ms ~308 ms ~319 ms
-	LFM2-VL 3B	3B	INT4 Q8 FP16	106 tok/s 69 tok/s 39 tok/s	~303 ms ~308 ms ~319 ms
-	Liquid LFM2-VL 3B D1	3B	INT4 Q8 FP16	106 tok/s 69 tok/s 39 tok/s	~303 ms ~308 ms ~319 ms
-	Qwen2.5-3B	3B	INT4 Q8 FP16	106 tok/s 69 tok/s 39 tok/s	~303 ms ~308 ms ~319 ms
-	ERNIE-4.5-21B-A3B	21B (3B active)	INT4	87 tok/s	~401 ms
-	ERNIE-4.5-21B-A3B-Base	21B (3B active)	INT4	87 tok/s	~401 ms
-	FrogNano 4B	4B	INT4 Q8 FP16	87 tok/s 55 tok/s 30 tok/s	~395 ms ~402 ms ~416 ms
-	Spark X2.5 4B	4B	INT4 Q8 FP16	87 tok/s 55 tok/s 30 tok/s	~395 ms ~402 ms ~416 ms
-	EmbeddingGemma 2	2B	INT4 Q8 FP16	83 tok/s 67 tok/s 45 tok/s	~217 ms ~220 ms ~228 ms
-	ERNIE-4.5-VL-28B-A3B	28B (3B active)	INT4	79 tok/s	~455 ms
-	ERNIE-4.5-VL-28B-A3B-Base	28B (3B active)	INT4	79 tok/s	~455 ms
-	Phi-2	2.7B	INT4 Q8 FP16	72 tok/s 56 tok/s 36 tok/s	~281 ms ~285 ms ~295 ms
-	MaLLaM-3B	3B	INT4 Q8 FP16	69 tok/s 52 tok/s 33 tok/s	~309 ms ~313 ms ~324 ms
-	SmolLM3 3B	3B	INT4 Q8 FP16	69 tok/s 52 tok/s 33 tok/s	~309 ms ~313 ms ~324 ms
-	Falcon-7B	7B	INT4 Q8 FP16	59 tok/s 35 tok/s 17 tok/s	~669 ms ~680 ms ~765 ms
-	Falcon3-7B	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
-	Hy-MT2 7B	7B	INT4 Q8 FP16	56 tok/s 34 tok/s 16 tok/s	~669 ms ~681 ms ~826 ms
-	Mistral-7B-v0.1	7.3B	INT4 Q8 FP16	54 tok/s 33 tok/s 16 tok/s	~696 ms ~708 ms ~860 ms
-	LensVLM 9B	9B	INT4 Q8	46 tok/s 27 tok/s	~854 ms ~869 ms
-	MiMo V2.6 Distill Qwen 9B	9B	INT4 Q8	46 tok/s 27 tok/s	~854 ms ~869 ms
-	ChatGLM3-6B-32K	6B	INT4 Q8 FP16	44 tok/s 31 tok/s 18 tok/s	~585 ms ~594 ms ~667 ms
-	Yi-6B	6B	INT4 Q8 FP16	44 tok/s 31 tok/s 18 tok/s	~585 ms ~594 ms ~667 ms
-	Falcon3-10B	10B	INT4 Q8	42 tok/s 25 tok/s	~945 ms ~962 ms
-	Falcon2-11B	11B	INT4 Q8	40 tok/s 23 tok/s	~1.0s ~1.1s
-	Hunyuan Lite	7B	INT4 Q8	40 tok/s 28 tok/s	~677 ms ~688 ms
-	Kimi-VL-A3B-Instruct	16B (3B active)	INT4	40 tok/s	~399 ms
-	OLMo 3 7B Base	7B	INT4 Q8	40 tok/s 28 tok/s	~677 ms ~688 ms
-	SEA-LION-7B	7.1B	INT4 Q8	39 tok/s 27 tok/s	~686 ms ~697 ms
-	SEA-LION-7B-Instruct	7.1B	INT4 Q8	39 tok/s 27 tok/s	~686 ms ~697 ms
-	Sahabat-AI-Llama3-8B-Instruct	8B	INT4 Q8	36 tok/s 25 tok/s	~769 ms ~781 ms
-	Typhoon-2-8B	8B	INT4 Q8	36 tok/s 25 tok/s	~769 ms ~781 ms
-	GLM-4-9B	9B	INT4 Q8	32 tok/s 21 tok/s	~863 ms ~951 ms
-	GLM-4-9B-Chat	9B	INT4 Q8	32 tok/s 21 tok/s	~863 ms ~951 ms
-	GLM-4-9B-Chat-1M	9B	INT4 Q8	32 tok/s 21 tok/s	~863 ms ~951 ms
-	Sahabat-AI-Gemma2-9B	9.2B	INT4 Q8	32 tok/s 21 tok/s	~881 ms ~971 ms
-	Sahabat-AI-Gemma2-9B-Instruct	9.2B	INT4 Q8	32 tok/s 21 tok/s	~881 ms ~971 ms
-	Yi-9B	9B	INT4 Q8	32 tok/s 21 tok/s	~863 ms ~951 ms
-	GreenMind-14B-R1	14B	INT4	22 tok/s	~1.3s
-	AREX-2	27B	INT4	15 tok/s	~2.9s
-	PPLX Decider v1 27B	27B	INT4	15 tok/s	~2.9s
-	PPLX Decider v1.1 27B	27B	INT4	15 tok/s	~2.9s

AMD RX 9060 XT (16GB)

VRAM Scaling by Context Length

Best Models to Run on AMD RX 9060 XT (16GB)

Inference Capacity

Fine-Tuning Capacity

Workload Recommendations

Frequently Asked Questions