What are the best models to run on NVIDIA RTX 4060 Ti (8GB)?

Based on model rankings and 8 GB usable VRAM: Top Overall: Qwen3.5-9B (INT4) ~42 tok/s. Top Coding: K2 Horizon 7B (INT4) ~51 tok/s. Optimal Daily Driver: Granite 4.2 8B (INT4) ~46 tok/s. Maximum Model Size: Gemma 3 12B (INT4) ~29 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 4060 Ti (8GB) (October 2026)

Rank	Model	Parameters	Precision	TPS (Approx.)	TTFT
#113	Qwen3.5-9B	9B	INT4	42 tok/s	~1.2s
#118	K2 Horizon 7B	7B	INT4	51 tok/s	~903 ms
#128	NVIDIA Nemotron 3 Nano 30B-A3B	3.5B (30B active)	INT4 Q8	13 tok/s 7 tok/s	~3.6s ~4.0s
#132	Qwen3.5-4B	4B	INT4 Q8	80 tok/s 50 tok/s	~532 ms ~539 ms
#146	Granite 4.2 8B	8B	INT4	46 tok/s	~1.0s
#150	Gemma 4 E4B	8B	INT4	46 tok/s	~1.0s
#169	OLMo 3 7B Think	7B	INT4	34 tok/s	~989 ms
#177	Gemma 3 12B	12B	INT4	29 tok/s	~1.8s
#182	Ministral 3 8B	8B	INT4	29 tok/s	~1.2s
#183	Qwen3.5-2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~285 ms ~289 ms ~297 ms
#184	MiniCPM5-2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~285 ms ~289 ms ~297 ms
#188	Gemma 4 E2B	5.1B	INT4 Q8	67 tok/s 39 tok/s	~668 ms ~735 ms
#189	Qwen3-4B	4B	INT4 Q8	80 tok/s 50 tok/s	~532 ms ~539 ms
#190	DeepSeek-R1 8B	8B	INT4	47 tok/s	~1.0s
#193	Ministral 3 3B	3B	INT4 Q8	63 tok/s 48 tok/s	~414 ms ~419 ms
#197	Gemma 2 9B	9B	INT4	42 tok/s	~1.2s
#198	Qwen2.5-7B	7B	INT4	51 tok/s	~903 ms
#199	Qwen3-8B	8B	INT4	46 tok/s	~1.0s
#200	Gemma 3 4B	4B	INT4 Q8	80 tok/s 50 tok/s	~532 ms ~539 ms
#201	DeepSeek-R1 7B	7B	INT4 Q8	53 tok/s 28 tok/s	~902 ms ~1.1s
#202	Llama 3.1 8B	8B	INT4	46 tok/s	~1.0s
#204	Qwen3-1.7B	1.7B	INT4 Q8 FP16	140 tok/s 98 tok/s 58 tok/s	~245 ms ~248 ms ~255 ms
#206	Gemma 3 270M	0.27B	INT4 Q8 FP16	256 tok/s 229 tok/s 183 tok/s	~44 ms ~45 ms ~46 ms
#208	Llama 3 8B	8B	INT4	46 tok/s	~1.0s
#209	Qwen3-0.6B	0.6B	INT4 Q8 FP16	236 tok/s 187 tok/s 128 tok/s	~90 ms ~91 ms ~94 ms
#210	Phi-4-Mini	3.8B	INT4 Q8	83 tok/s 53 tok/s	~507 ms ~514 ms
#212	Ministral-8B-2410	8B	INT4	46 tok/s	~1.0s
#213	DeepSeek-R1 1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~217 ms ~220 ms ~226 ms
#214	Llama 3.2 3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 34 tok/s	~408 ms ~413 ms ~461 ms
#215	Qwen3.5-0.8B	0.8B	INT4 Q8 FP16	209 tok/s 160 tok/s 105 tok/s	~120 ms ~121 ms ~124 ms
#216	Phi-3-mini	3.8B	INT4 Q8	83 tok/s 53 tok/s	~507 ms ~514 ms
#217	Gemma 3n E2B IT	6B (2B active)	INT4	60 tok/s	~338 ms
#218	Mistral-7B-Instruct-v0.1	7.3B	INT4	50 tok/s	~939 ms
#219	Llama 3.2 1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~150 ms ~152 ms ~156 ms
#220	Gemma 2 2B	2B	INT4 Q8 FP16	127 tok/s 87 tok/s 51 tok/s	~285 ms ~289 ms ~297 ms
#222	OLMo 3 7B Instruct	7B	INT4	34 tok/s	~989 ms
#224	Qwen2-7B	7B	INT4	51 tok/s	~903 ms
#225	Phi-3-small	7B	INT4	51 tok/s	~903 ms
#226	Gemma 3 1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~150 ms ~152 ms ~156 ms
#227	Mistral-7B-Instruct-v0.2	7.3B	INT4	50 tok/s	~939 ms
#229	Gemma 1 7B	7B	INT4	34 tok/s	~989 ms
#230	Gemma 1 2B	2B	INT4 Q8 FP16	137 tok/s 91 tok/s 52 tok/s	~284 ms ~288 ms ~296 ms
#231	ChatGLM3-6B	6B	INT4	38 tok/s	~854 ms
#232	ChatGLM2-6B	6B	INT4	38 tok/s	~854 ms
#233	ChatGLM-6B	6B	INT4	38 tok/s	~854 ms
Unranked Models (Sorted by Speed)
-	Qwen2-0.5B	0.5B	INT4 Q8 FP16	252 tok/s 204 tok/s 144 tok/s	~76 ms ~77 ms ~79 ms
-	Qwen2.5-0.5B	0.5B	INT4 Q8 FP16	252 tok/s 204 tok/s 144 tok/s	~76 ms ~77 ms ~79 ms
-	ERNIE-4.5-0.3B	0.3B	INT4 Q8 FP16	245 tok/s 218 tok/s 172 tok/s	~49 ms ~49 ms ~50 ms
-	ERNIE-4.5-0.3B-Base	0.3B	INT4 Q8 FP16	245 tok/s 218 tok/s 172 tok/s	~49 ms ~49 ms ~50 ms
-	Falcon-1B	1B	INT4 Q8 FP16	199 tok/s 145 tok/s 91 tok/s	~150 ms ~152 ms ~156 ms
-	Falcon3-1B	1B	INT4 Q8 FP16	186 tok/s 139 tok/s 89 tok/s	~150 ms ~152 ms ~156 ms
-	Qwen2-1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~217 ms ~220 ms ~226 ms
-	Qwen2.5-1.5B	1.5B	INT4 Q8 FP16	151 tok/s 107 tok/s 65 tok/s	~217 ms ~220 ms ~226 ms
-	Spark X2.5 1.7B	1.7B	INT4 Q8 FP16	140 tok/s 98 tok/s 58 tok/s	~245 ms ~248 ms ~255 ms
-	Hy-MT2 1.8B	1.8B	INT4 Q8 FP16	136 tok/s 94 tok/s 55 tok/s	~257 ms ~261 ms ~268 ms
-	Falcon-3B	3B	INT4 Q8 FP16	105 tok/s 66 tok/s 35 tok/s	~407 ms ~413 ms ~460 ms
-	CroissantLLM Base	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~194 ms ~196 ms ~201 ms
-	Phi-1	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~194 ms ~196 ms ~201 ms
-	Phi-1.5	1.3B	INT4 Q8 FP16	103 tok/s 84 tok/s 58 tok/s	~194 ms ~196 ms ~201 ms
-	DeepSeek-R1 3B	3B	INT4 Q8 FP16	100 tok/s 64 tok/s 34 tok/s	~408 ms ~413 ms ~461 ms
-	Falcon3-3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 34 tok/s	~408 ms ~413 ms ~461 ms
-	Qwen2.5-3B	3B	INT4 Q8 FP16	98 tok/s 64 tok/s 34 tok/s	~408 ms ~413 ms ~461 ms
-	Spark X2.5 4B	4B	INT4 Q8	80 tok/s 50 tok/s	~532 ms ~539 ms
-	Phi-2	2.7B	INT4 Q8 FP16	66 tok/s 51 tok/s 29 tok/s	~377 ms ~381 ms ~458 ms
-	MaLLaM-3B	3B	INT4 Q8	63 tok/s 48 tok/s	~414 ms ~419 ms
-	SmolLM3 3B	3B	INT4 Q8	63 tok/s 48 tok/s	~414 ms ~419 ms
-	Falcon-7B	7B	INT4 Q8	54 tok/s 29 tok/s	~902 ms ~1.1s
-	Falcon3-7B	7B	INT4	51 tok/s	~903 ms
-	Hy-MT2 7B	7B	INT4	51 tok/s	~903 ms
-	Mistral-7B-v0.1	7.3B	INT4	50 tok/s	~939 ms
-	LensVLM 9B	9B	INT4	42 tok/s	~1.2s
-	MiMo V2.6 Distill Qwen 9B	9B	INT4	42 tok/s	~1.2s
-	ChatGLM3-6B-32K	6B	INT4	38 tok/s	~854 ms
-	Yi-6B	6B	INT4	38 tok/s	~854 ms
-	Falcon3-10B	10B	INT4	36 tok/s	~1.4s
-	Falcon2-11B	11B	INT4	35 tok/s	~1.5s
-	Hunyuan Lite	7B	INT4	34 tok/s	~989 ms
-	OLMo 3 7B Base	7B	INT4	34 tok/s	~989 ms
-	SEA-LION-7B	7.1B	INT4	34 tok/s	~1.0s
-	SEA-LION-7B-Instruct	7.1B	INT4	34 tok/s	~1.0s
-	Sahabat-AI-Llama3-8B-Instruct	8B	INT4	29 tok/s	~1.2s
-	Typhoon-2-8B	8B	INT4	29 tok/s	~1.2s

NVIDIA RTX 4060 Ti (8GB)

VRAM Scaling by Context Length

Best Models to Run on NVIDIA RTX 4060 Ti (8GB)

Inference Capacity

Fine-Tuning Capacity

Workload Recommendations

Frequently Asked Questions