RTX 4080 Super 16GB — local LLM fit & speed

NVIDIA · 16 GB GDDR6X · 736 GB/s · 320 W · MSRP $999 · ASUS ProArt GeForce RTX 4080 Super OC (affiliate) · More listings

The RTX 4080 Super 16GB is a NVIDIA accelerator with 16 GB GDDR6X and about 736 GB/s peak memory bandwidth. AI Hashrate estimates local LLM decode (batch 1) at 8K context: fit counts weights + KV cache + 1 GB runtime overhead — not weights alone. At Q4, about 25 curated models fit fully on this card; at FP16, about 10 fit. Top Q4 speeds: Qwen3-0.6B ≈ 585.5 tok/s (estimated, fits); gemma-3-1b-it ≈ 468.4 tok/s (estimated, fits); Llama-3.2-1B-Instruct ≈ 390.3 tok/s (estimated, fits); Qwen3-1.7B ≈ 234.2 tok/s (estimated, fits); gpt-oss-20b ≈ 186.5 tok/s (measured, fits). Relative ranking is more reliable than absolute tok/s. See methodology for the bandwidth formula and measured-anchor policy. Methodology.

Table: decode speed estimates (Q4 / FP16). Measured rows override estimates. MSRP is list price, not live retail. Context default 8K.

Model	Params (B)	Quant	tok/s	Fits?
Qwen3-0.6B	0.8	Q4	585.5 est.	Yes
gemma-3-1b-it	1.0	Q4	468.4 est.	Yes
Llama-3.2-1B-Instruct	1.2	Q4	390.3 est.	Yes
Qwen3-30B-A3B	30.5	Q4	338.0 est.	No
Qwen3-Coder-30B-A3B-Instruct	30.5	Q4	338.0 est.	No
Qwen3-30B-A3B-Instruct-2507	30.5	Q4	338.0 est.	No
Qwen3-1.7B	2.0	Q4	234.2 est.	Yes
gpt-oss-20b	21.5	Q4	186.5 measured (llama.cpp-disc-15396)	Yes
Qwen3-0.6B	0.8	FP16	161.0 est.	Yes
Llama-3.2-3B-Instruct	3.2	Q4	146.4 est.	Yes
gemma-3-1b-it	1.0	FP16	128.8 est.	Yes
Phi-4-mini-instruct	3.8	Q4	123.3 est.	Yes
Phi-4-mini-reasoning	3.8	Q4	123.3 est.	Yes
gemma-3-4b-it	4.3	Q4	120.1 est.	Yes
Qwen3-4B-Instruct-2507	4.0	Q4	117.1 est.	Yes
Qwen3-4B-Thinking-2507	4.0	Q4	117.1 est.	Yes
Llama-3.2-1B-Instruct	1.2	FP16	107.3 est.	Yes
GLM-4.7-Flash	31.2	Q4	67.5 est.	No
Mistral-7B-Instruct-v0.3	7.2	Q4	65.0 measured (llama-cpp-bench)	Yes
Qwen3-1.7B	2.0	FP16	64.4 est.	Yes
Qwen2.5-7B-Instruct	7.6	Q4	61.6 est.	Yes
Qwen2.5-Coder-7B-Instruct	7.6	Q4	61.6 est.	Yes
DeepSeek-R1-Distill-Qwen-7B	7.6	Q4	61.6 est.	Yes
Llama-3.1-8B-Instruct	8.0	Q4	58.0 measured (llama-cpp-bench)	Yes
Qwen3-8B	8.2	Q4	57.1 est.	Yes
DeepSeek-R1-0528-Qwen3-8B	8.2	Q4	57.1 est.	Yes
Qwen3-Coder-Next	79.7	Q4	55.3 est.	No
Qwen3-Next-80B-A3B-Instruct	81.3	Q4	53.2 est.	No
Mistral-7B-Instruct-v0.3	7.2	FP16	46.1 est.	No
Devstral-Small-2507	23.6	Q4	41.5 est.	No
Llama-3.2-3B-Instruct	3.2	FP16	40.2 est.	Yes
Qwen2.5-7B-Instruct	7.6	FP16	39.5 est.	No
Qwen2.5-Coder-7B-Instruct	7.6	FP16	39.5 est.	No
DeepSeek-R1-Distill-Qwen-7B	7.6	FP16	39.5 est.	No
Mistral-Small-3.2-24B-Instruct-2506	24.0	Q4	39.5 est.	No
Magistral-Small-2509	24.0	Q4	39.5 est.	No
gemma-3-12b-it	12.2	Q4	39.0 est.	Yes
Llama-3.1-8B-Instruct	8.0	FP16	34.0 est.	No
Phi-4-mini-instruct	3.8	FP16	33.9 est.	Yes
Phi-4-mini-reasoning	3.8	FP16	33.9 est.	Yes
gemma-3-4b-it	4.3	FP16	33.0 est.	Yes
Qwen3-4B-Instruct-2507	4.0	FP16	32.2 est.	Yes
Qwen3-4B-Thinking-2507	4.0	FP16	32.2 est.	Yes
Phi-4	14.7	Q4	31.9 est.	Yes
Phi-4-reasoning	14.7	Q4	31.9 est.	Yes
Qwen3-8B	8.2	FP16	31.7 est.	No
DeepSeek-R1-0528-Qwen3-8B	8.2	FP16	31.7 est.	No
Qwen3-14B	14.8	Q4	31.6 est.	Yes
Qwen2.5-14B-Instruct	14.8	Q4	31.6 est.	Yes
Qwen2.5-Coder-14B-Instruct	14.8	Q4	31.6 est.	Yes
DeepSeek-R1-Distill-Qwen-14B	14.8	Q4	31.6 est.	Yes
Mixtral-8x7B-Instruct-v0.1	46.7	Q4	31.0 est.	No
gemma-3-27b-it	27.4	Q4	27.2 est.	No
Qwen3-32B	32.8	Q4	15.9 est.	No
Qwen2.5-32B-Instruct	32.8	Q4	15.9 est.	No
Qwen2.5-Coder-32B-Instruct	32.8	Q4	15.9 est.	No
DeepSeek-R1-Distill-Qwen-32B	32.8	Q4	15.9 est.	No
gpt-oss-20b	21.5	FP16	13.1 est.	No
gemma-3-12b-it	12.2	FP16	10.2 est.	No
Qwen3-30B-A3B	30.5	FP16	8.0 est.	No
Qwen3-Coder-30B-A3B-Instruct	30.5	FP16	8.0 est.	No
Qwen3-30B-A3B-Instruct-2507	30.5	FP16	8.0 est.	No
Phi-4	14.7	FP16	5.8 est.	No
Phi-4-reasoning	14.7	FP16	5.8 est.	No
Qwen3-14B	14.8	FP16	5.7 est.	No
Qwen2.5-14B-Instruct	14.8	FP16	5.7 est.	No
Qwen2.5-Coder-14B-Instruct	14.8	FP16	5.7 est.	No
DeepSeek-R1-Distill-Qwen-14B	14.8	FP16	5.7 est.	No
Llama-3.1-70B-Instruct	70.6	Q4	1.7 est.	No
Llama-3.3-70B-Instruct	70.6	Q4	1.7 est.	No
DeepSeek-R1-Distill-Llama-70B	70.6	Q4	1.7 est.	No
Qwen2.5-72B-Instruct	72.7	Q4	1.5 est.	No
Mistral-Small-3.2-24B-Instruct-2506	24.0	FP16	1.4 est.	No
Magistral-Small-2509	24.0	FP16	1.4 est.	No
Devstral-Small-2507	23.6	FP16	1.4 est.	No
gemma-3-27b-it	27.4	FP16	0.9 est.	No