What are the best models to run on NVIDIA RTX 5080 (16GB)?

Based on model rankings and 16 GB usable VRAM: Top Overall: Qwen 3.8 27B (INT4) ~42 tok/s. Top Coding: Qwen3.5-27B (INT4) ~42 tok/s. Optimal Daily Driver: Phi-4 Reasoning Plus (INT4) ~60 tok/s. Maximum Model Size: ERNIE-4.5-VL-28B-A3B (INT4) ~161 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 NVIDIA RTX 5080 (16GB) (October 2026)

Rank	Model	Parameters	Precision	TPS (Approx.)	TTFT
#33	Qwen 3.8 27B	27B	INT4	42 tok/s	~1.6s
#83	Qwen3.5-27B	27B	INT4	42 tok/s	~1.6s
#100	Phi-4 Reasoning Plus	14B	INT4	60 tok/s	~706 ms
#101	MiMo V2 Flash	15B (309B active)	INT4	4 tok/s	~15.7s
#108	Gemma 4 26B A4B	25.2B (3.8B active)	INT4	157 tok/s	~263 ms
#109	Gemma 4 12B	11.95B	INT4 Q8	67 tok/s 42 tok/s	~605 ms ~716 ms
#113	Qwen3.5-9B	9B	INT4 Q8	110 tok/s 71 tok/s	~457 ms ~462 ms
#118	K2 Horizon 7B	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
#128	NVIDIA Nemotron 3 Nano 30B-A3B	3.5B (30B active)	INT4 Q8 FP16	40 tok/s 24 tok/s 13 tok/s	~1.4s ~1.4s ~1.5s
#132	Qwen3.5-4B	4B	INT4 Q8 FP16	177 tok/s 127 tok/s 78 tok/s	~213 ms ~216 ms ~220 ms
#136	GPT-OSS 20B	21B (3.6B active)	INT4	81 tok/s	~272 ms
#146	Granite 4.2 8B	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#150	Gemma 4 E4B	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#158	GLM-4V	9B	INT4 Q8	83 tok/s 56 tok/s	~460 ms ~504 ms
#159	Gemma 3 27B	27B	INT4	42 tok/s	~1.6s
#160	Phi-4	14B	INT4 Q8	79 tok/s 44 tok/s	~702 ms ~831 ms
#169	OLMo 3 7B Think	7B	INT4 Q8	98 tok/s 72 tok/s	~362 ms ~365 ms
#172	DeepSeek-R1 14B	14B	INT4 Q8	81 tok/s 45 tok/s	~701 ms ~831 ms
#173	Ministral 3 14B	14B	INT4	60 tok/s	~706 ms
#177	Gemma 3 12B	12B	INT4 Q8	89 tok/s 53 tok/s	~604 ms ~662 ms
#181	Mistral-Small-2501	24B	INT4	49 tok/s	~1.3s
#182	Ministral 3 8B	8B	INT4 Q8	90 tok/s 65 tok/s	~410 ms ~414 ms
#183	Qwen3.5-2B	2B	INT4 Q8 FP16	235 tok/s 187 tok/s 128 tok/s	~117 ms ~118 ms ~120 ms
#184	MiniCPM5-2B	2B	INT4 Q8 FP16	235 tok/s 187 tok/s 128 tok/s	~117 ms ~118 ms ~120 ms
#185	Qwen2.5-14B	14B	INT4 Q8	79 tok/s 44 tok/s	~702 ms ~831 ms
#186	Gemma 2 27B	27B	INT4	42 tok/s	~1.6s
#188	Gemma 4 E2B	5.1B	INT4 Q8 FP16	156 tok/s 108 tok/s 65 tok/s	~267 ms ~270 ms ~276 ms
#189	Qwen3-4B	4B	INT4 Q8 FP16	177 tok/s 127 tok/s 78 tok/s	~213 ms ~216 ms ~220 ms
#190	DeepSeek-R1 8B	8B	INT4 Q8	121 tok/s 79 tok/s	~407 ms ~412 ms
#193	Ministral 3 3B	3B	INT4 Q8 FP16	150 tok/s 122 tok/s 84 tok/s	~166 ms ~168 ms ~172 ms
#195	Qwen3-14B	14B	INT4 Q8	79 tok/s 44 tok/s	~702 ms ~831 ms
#197	Gemma 2 9B	9B	INT4 Q8	110 tok/s 71 tok/s	~457 ms ~462 ms
#198	Qwen2.5-7B	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
#199	Qwen3-8B	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#200	Gemma 3 4B	4B	INT4 Q8 FP16	177 tok/s 127 tok/s 78 tok/s	~213 ms ~216 ms ~220 ms
#201	DeepSeek-R1 7B	7B	INT4 Q8 FP16	132 tok/s 87 tok/s 47 tok/s	~359 ms ~363 ms ~402 ms
#202	Llama 3.1 8B	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#204	Qwen3-1.7B	1.7B	INT4 Q8 FP16	248 tok/s 201 tok/s 142 tok/s	~101 ms ~102 ms ~104 ms
#206	Gemma 3 270M	0.27B	INT4 Q8 FP16	327 tok/s 313 tok/s 283 tok/s	~22 ms ~22 ms ~22 ms
#208	Llama 3 8B	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#209	Qwen3-0.6B	0.6B	INT4 Q8 FP16	316 tok/s 286 tok/s 237 tok/s	~40 ms ~40 ms ~41 ms
#210	Phi-4-Mini	3.8B	INT4 Q8 FP16	182 tok/s 132 tok/s 82 tok/s	~204 ms ~206 ms ~210 ms
#212	Ministral-8B-2410	8B	INT4 Q8	119 tok/s 78 tok/s	~408 ms ~412 ms
#213	DeepSeek-R1 1.5B	1.5B	INT4 Q8 FP16	258 tok/s 213 tok/s 153 tok/s	~90 ms ~91 ms ~93 ms
#214	Llama 3.2 3B	3B	INT4 Q8 FP16	202 tok/s 151 tok/s 97 tok/s	~165 ms ~166 ms ~170 ms
#215	Qwen3.5-0.8B	0.8B	INT4 Q8 FP16	300 tok/s 265 tok/s 211 tok/s	~52 ms ~52 ms ~53 ms
#216	Phi-3-mini	3.8B	INT4 Q8 FP16	182 tok/s 132 tok/s 82 tok/s	~204 ms ~206 ms ~210 ms
#217	Gemma 3n E2B IT	6B (2B active)	INT4 Q8 FP16	150 tok/s 132 tok/s 95 tok/s	~126 ms ~127 ms ~140 ms
#218	Mistral-7B-Instruct-v0.1	7.3B	INT4 Q8 FP16	126 tok/s 83 tok/s 43 tok/s	~373 ms ~377 ms ~452 ms
#219	Llama 3.2 1B	1B	INT4 Q8 FP16	286 tok/s 247 tok/s 190 tok/s	~64 ms ~64 ms ~65 ms
#220	Gemma 2 2B	2B	INT4 Q8 FP16	235 tok/s 187 tok/s 128 tok/s	~117 ms ~118 ms ~120 ms
#221	Phi-3-medium	14B	INT4 Q8	79 tok/s 44 tok/s	~702 ms ~831 ms
#222	OLMo 3 7B Instruct	7B	INT4 Q8	98 tok/s 72 tok/s	~362 ms ~365 ms
#224	Qwen2-7B	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
#225	Phi-3-small	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
#226	Gemma 3 1B	1B	INT4 Q8 FP16	286 tok/s 247 tok/s 190 tok/s	~64 ms ~64 ms ~65 ms
#227	Mistral-7B-Instruct-v0.2	7.3B	INT4 Q8 FP16	126 tok/s 83 tok/s 43 tok/s	~373 ms ~377 ms ~452 ms
#229	Gemma 1 7B	7B	INT4 Q8	98 tok/s 72 tok/s	~362 ms ~365 ms
#230	Gemma 1 2B	2B	INT4 Q8 FP16	245 tok/s 192 tok/s 131 tok/s	~116 ms ~117 ms ~120 ms
#231	ChatGLM3-6B	6B	INT4 Q8 FP16	107 tok/s 80 tok/s 48 tok/s	~313 ms ~316 ms ~350 ms
#232	ChatGLM2-6B	6B	INT4 Q8 FP16	107 tok/s 80 tok/s 48 tok/s	~313 ms ~316 ms ~350 ms
#233	ChatGLM-6B	6B	INT4 Q8 FP16	107 tok/s 80 tok/s 48 tok/s	~313 ms ~316 ms ~350 ms
Unranked Models (Sorted by Speed)
-	Qwen2-0.5B	0.5B	INT4 Q8 FP16	324 tok/s 298 tok/s 252 tok/s	~34 ms ~35 ms ~35 ms
-	Qwen2.5-0.5B	0.5B	INT4 Q8 FP16	324 tok/s 298 tok/s 252 tok/s	~34 ms ~35 ms ~35 ms
-	ERNIE-4.5-0.3B	0.3B	INT4 Q8 FP16	321 tok/s 306 tok/s 275 tok/s	~24 ms ~24 ms ~24 ms
-	ERNIE-4.5-0.3B-Base	0.3B	INT4 Q8 FP16	321 tok/s 306 tok/s 275 tok/s	~24 ms ~24 ms ~24 ms
-	Falcon-1B	1B	INT4 Q8 FP16	294 tok/s 253 tok/s 193 tok/s	~64 ms ~64 ms ~65 ms
-	Falcon3-1B	1B	INT4 Q8 FP16	286 tok/s 247 tok/s 190 tok/s	~64 ms ~64 ms ~65 ms
-	Qwen2-1.5B	1.5B	INT4 Q8 FP16	258 tok/s 213 tok/s 153 tok/s	~90 ms ~91 ms ~93 ms
-	Qwen2.5-1.5B	1.5B	INT4 Q8 FP16	258 tok/s 213 tok/s 153 tok/s	~90 ms ~91 ms ~93 ms
-	Spark X2.5 1.7B	1.7B	INT4 Q8 FP16	248 tok/s 201 tok/s 142 tok/s	~101 ms ~102 ms ~104 ms
-	Hy-MT2 1.8B	1.8B	INT4 Q8 FP16	244 tok/s 196 tok/s 137 tok/s	~106 ms ~107 ms ~109 ms
-	Falcon-3B	3B	INT4 Q8 FP16	210 tok/s 156 tok/s 99 tok/s	~165 ms ~166 ms ~170 ms
-	CroissantLLM Base	1.3B	INT4 Q8 FP16	208 tok/s 184 tok/s 142 tok/s	~80 ms ~81 ms ~82 ms
-	Phi-1	1.3B	INT4 Q8 FP16	208 tok/s 184 tok/s 142 tok/s	~80 ms ~81 ms ~82 ms
-	Phi-1.5	1.3B	INT4 Q8 FP16	208 tok/s 184 tok/s 142 tok/s	~80 ms ~81 ms ~82 ms
-	DeepSeek-R1 3B	3B	INT4 Q8 FP16	204 tok/s 152 tok/s 98 tok/s	~165 ms ~166 ms ~170 ms
-	Falcon3-3B	3B	INT4 Q8 FP16	202 tok/s 151 tok/s 97 tok/s	~165 ms ~166 ms ~170 ms
-	Qwen2.5-3B	3B	INT4 Q8 FP16	202 tok/s 151 tok/s 97 tok/s	~165 ms ~166 ms ~170 ms
-	Spark X2.5 4B	4B	INT4 Q8 FP16	177 tok/s 127 tok/s 78 tok/s	~213 ms ~216 ms ~220 ms
-	ERNIE-4.5-21B-A3B	21B (3B active)	INT4	174 tok/s	~216 ms
-	ERNIE-4.5-21B-A3B-Base	21B (3B active)	INT4	174 tok/s	~216 ms
-	ERNIE-4.5-VL-28B-A3B	28B (3B active)	INT4	161 tok/s	~245 ms
-	ERNIE-4.5-VL-28B-A3B-Base	28B (3B active)	INT4	161 tok/s	~245 ms
-	Phi-2	2.7B	INT4 Q8 FP16	156 tok/s 129 tok/s 90 tok/s	~152 ms ~153 ms ~157 ms
-	MaLLaM-3B	3B	INT4 Q8 FP16	150 tok/s 122 tok/s 84 tok/s	~166 ms ~168 ms ~172 ms
-	SmolLM3 3B	3B	INT4 Q8 FP16	150 tok/s 122 tok/s 84 tok/s	~166 ms ~168 ms ~172 ms
-	Falcon-7B	7B	INT4 Q8 FP16	134 tok/s 88 tok/s 48 tok/s	~359 ms ~363 ms ~402 ms
-	Falcon3-7B	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
-	Hy-MT2 7B	7B	INT4 Q8 FP16	130 tok/s 86 tok/s 45 tok/s	~359 ms ~363 ms ~434 ms
-	Mistral-7B-v0.1	7.3B	INT4 Q8 FP16	126 tok/s 83 tok/s 43 tok/s	~373 ms ~377 ms ~452 ms
-	LensVLM 9B	9B	INT4 Q8	110 tok/s 71 tok/s	~457 ms ~462 ms
-	MiMo V2.6 Distill Qwen 9B	9B	INT4 Q8	110 tok/s 71 tok/s	~457 ms ~462 ms
-	ChatGLM3-6B-32K	6B	INT4 Q8 FP16	107 tok/s 80 tok/s 48 tok/s	~313 ms ~316 ms ~350 ms
-	Yi-6B	6B	INT4 Q8 FP16	107 tok/s 80 tok/s 48 tok/s	~313 ms ~316 ms ~350 ms
-	Falcon3-10B	10B	INT4 Q8	102 tok/s 65 tok/s	~506 ms ~511 ms
-	Falcon2-11B	11B	INT4 Q8	99 tok/s 61 tok/s	~554 ms ~560 ms
-	Hunyuan Lite	7B	INT4 Q8	98 tok/s 72 tok/s	~362 ms ~365 ms
-	OLMo 3 7B Base	7B	INT4 Q8	98 tok/s 72 tok/s	~362 ms ~365 ms
-	Kimi-VL-A3B-Instruct	16B (3B active)	INT4	97 tok/s	~212 ms
-	SEA-LION-7B	7.1B	INT4 Q8	97 tok/s 71 tok/s	~366 ms ~370 ms
-	SEA-LION-7B-Instruct	7.1B	INT4 Q8	97 tok/s 71 tok/s	~366 ms ~370 ms
-	Sahabat-AI-Llama3-8B-Instruct	8B	INT4 Q8	90 tok/s 65 tok/s	~410 ms ~414 ms
-	Typhoon-2-8B	8B	INT4 Q8	90 tok/s 65 tok/s	~410 ms ~414 ms
-	GLM-4-9B	9B	INT4 Q8	83 tok/s 56 tok/s	~460 ms ~504 ms
-	GLM-4-9B-Chat	9B	INT4 Q8	83 tok/s 56 tok/s	~460 ms ~504 ms
-	GLM-4-9B-Chat-1M	9B	INT4 Q8	83 tok/s 56 tok/s	~460 ms ~504 ms
-	Yi-9B	9B	INT4 Q8	83 tok/s 56 tok/s	~460 ms ~504 ms
-	Sahabat-AI-Gemma2-9B	9.2B	INT4 Q8	82 tok/s 56 tok/s	~469 ms ~514 ms
-	Sahabat-AI-Gemma2-9B-Instruct	9.2B	INT4 Q8	82 tok/s 56 tok/s	~469 ms ~514 ms
-	GreenMind-14B-R1	14B	INT4	60 tok/s	~706 ms

NVIDIA RTX 5080 (16GB)

VRAM Scaling by Context Length

Best Models to Run on NVIDIA RTX 5080 (16GB)

Inference Capacity

Fine-Tuning Capacity

Workload Recommendations

Frequently Asked Questions