NVIDIA · Ada

What runs on a RTX 4070 Ti Super

16 GB at 672 GB/s. The largest model that fits at Q4_K_M is gpt-oss 20B-A3.6B, at roughly 134 tokens per second.

Model

Hardware

Quantisation

Context 8K

KV cache precision

Runtime

Batch

GPUs

Monitor plugged into this GPU

Runs comfortably

6.4 GB of 14.2 GB · 45%

014 GB

Weights 4.6 GB

KV cache 1.0 GB

Runtime overhead 0.8 GB

Llama 3.1 8B at Q4_K_M leaves 7.8 GB spare on a RTX 4070 Ti Super. There is room to raise the context length or move up a quantisation level.

Generation77.6tok/s

Prompt processing2301tok/s

Max context70Ktokens

KV per 1K tokens0GB

Every quantisation of Llama 3.1 8B on a RTX 4070 Ti Super

Highlighted row is the highest quality that still fits at 8K context.

Quantisation	bpw	Weights	Total	Verdict	Max ctx	tok/s
FP16 / BF16	16.00	15.0 GB	16.8 GB	Won't fit	—	9.34
INT8 / W8A8	8.50	7.7 GB	9.5 GB	Runs comfortably	46K	48.6
Q8_0 (GGUF)	8.50	7.7 GB	9.5 GB	Runs comfortably	46K	48.6
FP8 (E4M3)	8.00	7.5 GB	9.3 GB	Runs comfortably	48K	50.0
Q6_K	6.56	6.1 GB	8.0 GB	Runs comfortably	58K	60.1
AWQ 4-bit	4.25	5.4 GB	7.2 GB	Runs comfortably	64K	67.3
GPTQ 4-bit	4.25	5.4 GB	7.2 GB	Runs comfortably	64K	67.3
MXFP4	4.25	5.4 GB	7.2 GB	Runs comfortably	64K	67.3
Q5_K_M	5.67	5.3 GB	7.1 GB	Runs comfortably	65K	68.6
Q5_K_S	5.52	5.2 GB	7.0 GB	Runs comfortably	66K	70.2
Q4_K_M	4.85	4.6 GB	6.4 GB	Runs comfortably	70K	77.6
Q4_K_S	4.58	4.4 GB	6.2 GB	Runs comfortably	72K	81.1
Q4_0	4.55	4.4 GB	6.2 GB	Runs comfortably	72K	81.5
IQ4_XS	4.25	4.1 GB	5.9 GB	Runs comfortably	74K	85.7
Q3_K_M	3.91	3.8 GB	5.7 GB	Runs comfortably	77K	91.0
IQ3_M	3.70	3.7 GB	5.5 GB	Runs comfortably	78K	94.7
IQ3_XXS	3.06	3.2 GB	5.0 GB	Runs comfortably	82K	108
Q2_K	2.63	2.8 GB	4.6 GB	Runs comfortably	85K	119
IQ2_XXS	2.06	2.3 GB	4.2 GB	Runs comfortably	89K	138
IQ1_M	1.75	2.1 GB	3.9 GB	Runs comfortably	91K	151

Q4_K_M at 8K context. Bandwidth sets the speed; capacity sets the ceiling.

Model	Params	Weights	Verdict	Max ctx	tok/s
DeepSeek-R1 671B-A37B	671B	379.0 GB	Won't fit	—	1.84
Qwen3 235B-A22B	235.1B	132.8 GB	Won't fit	—	3.05
gpt-oss 120B-A5.1B	116.8B	66.0 GB	Won't fit	—	14.3
Llama 4 Scout 109B-A17B	109B	61.7 GB	Won't fit	—	4.42
Qwen2.5 72B	72.7B	41.2 GB	Won't fit	—	1.18
Llama 3.1 70B	70.5B	40.0 GB	Won't fit	—	1.22
Mixtral 8x7B	46.7B	26.4 GB	Won't fit	—	8.38
Command R 35B	35.0B	20.1 GB	Won't fit	—	1.77
Yi-1.5 34B	34.4B	19.5 GB	Won't fit	—	3.96
Qwen3 32B	32.8B	18.6 GB	Won't fit	—	4.34
Qwen2.5-Coder 32B	32.8B	18.6 GB	Won't fit	—	4.34
DeepSeek-R1-Distill-Qwen 32B	32.8B	18.6 GB	Won't fit	—	4.34
Qwen3 30B-A3B	30.5B	17.3 GB	Won't fit	—	40.2
Gemma 3 27B	27.4B	15.6 GB	Won't fit	—	7.81
Mistral Small 24B	23.6B	13.4 GB	Won't fit	—	14.7
gpt-oss 20B-A3.6B	20.9B	11.9 GB	Fits, but tight	33K	134
Qwen3 14B	14.8B	8.5 GB	Runs comfortably	32K	43.9
Phi-4 14B	14.7B	8.4 GB	Runs comfortably	16K	43.5
Mistral NeMo 12B	12.3B	7.0 GB	Runs comfortably	41K	52.1
Gemma 3 12B	12.2B	7.0 GB	Runs comfortably	98K	52.7
GLM-4 9B	9.4B	5.4 GB	Runs comfortably	128K	71.3
Qwen3 8B	8.2B	4.7 GB	Runs comfortably	62K	75.1
Llama 3.1 8B	8.0B	4.6 GB	Runs comfortably	70K	77.6
DeepSeek-R1-Distill-Llama 8B	8.0B	4.6 GB	Runs comfortably	70K	77.6
Qwen2.5 7B	7.6B	4.4 GB	Runs comfortably	128K	86.3
Mistral 7B v0.3	7.3B	4.1 GB	Runs comfortably	32K	85.9
Gemma 3 4B	4.3B	2.5 GB	Runs comfortably	128K	146
Qwen3 4B	4.0B	2.3 GB	Runs comfortably	79K	137
Llama 3.2 3B	3.2B	1.8 GB	Runs comfortably	106K	172
Llama 3.2 1B	1.2B	0.7 GB	Runs comfortably	128K	456