Qwen2.5-72B-Instruct

Alibaba · 72.7B · ctx 128K · Apache 2.0 · HuggingFace

Qwen2.5-72B-Instruct (Alibaba) is a 72.7B parameters open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 55.53 GB VRAM for Q4 (weights 39.99 GB + KV 14.54 GB + 1 GB overhead) and 160.94 GB for FP16. About 11 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 70.0 tok/s (estimated); MI300X 192GB ≈ 46.4 tok/s (estimated); H200 141GB SXM5 ≈ 42.0 tok/s (estimated); H100 80GB SXM5 ≈ 29.3 tok/s (estimated); Gaudi 3 128GB ≈ 28.7 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.

49 hardware configs — sorted by tok/s. Default fit context 8K.

Hardware	VRAM	Quant	tok/s	Fits?	tok/s/$
B200 192GB SXM	192	Q4	70.0 est.	Yes	0.002
MI300X 192GB	192	Q4	46.4 est.	Yes	0.003
H200 141GB SXM5	141	Q4	42.0 est.	Yes	0.001
H100 80GB SXM5	80	Q4	29.3 est.	Yes	0.001
Gaudi 3 128GB	128	Q4	28.7 est.	Yes	0.002
H200 141GB SXM5	141	FP16	25.3 est.	No	0.001
B200 192GB SXM	192	FP16	19.3 est.	Yes	0.0
RTX 6000 Ada 48GB	48	Q4	17.9 est.	No	0.003
A100 80GB SXM4	80	Q4	16.47 measured	Yes	0.001
L40S 48GB	48	Q4	16.1 est.	No	0.002
RTX 5090 32GB	32	Q4	14.9 est.	No	0.007
RTX A6000 48GB	48	Q4	14.4 est.	No	0.003
Gaudi 3 128GB	128	FP16	14.3 est.	No	0.001
MI300X 192GB	192	FP16	12.8 est.	Yes	0.001
MacBook Pro M4 Max 128GB	128	Q4	11.0 measured	Yes	0.003
Mac Studio M3 Ultra 96GB	96	Q4	7.2 est.	Yes	0.002
Mac Studio M2 Ultra 192GB	192	Q4	7.0 est.	Yes	0.001
H100 80GB SXM5	80	FP16	5.7 est.	No	0.0
RTX 4090 24GB	24	Q4	4.7 est.	No	0.003
RTX 4090D 24GB (China)	24	Q4	4.7 est.	No	0.003
RTX 3090 Ti 24GB	24	Q4	4.7 est.	No	0.002
RX 7900 XTX 24GB	24	Q4	4.5 est.	No	0.005
RTX 3090 24GB	24	Q4	4.4 est.	No	0.003
A100 80GB SXM4	80	FP16	3.5 est.	No	0.0
MacBook Pro M3 Max 128GB	128	Q4	3.5 est.	Yes	0.001
RX 7900 XT 20GB	20	Q4	2.6 est.	No	0.003
MacBook Pro M4 Max 128GB	128	FP16	2.4 est.	No	0.001
Mac mini M4 Pro 64GB	64	Q4	2.4 est.	Yes	0.001
RTX 5080 16GB	16	Q4	2.0 est.	No	0.002
Mac Studio M3 Ultra 96GB	96	FP16	2.0 est.	No	0.001
RTX 5070 Ti 16GB	16	Q4	1.9 est.	No	0.003
Mac Studio M2 Ultra 192GB	192	FP16	1.9 est.	Yes	0.0
MacBook Pro M3 Max 128GB	128	FP16	1.7 est.	No	0.0
RTX 4080 Super 16GB	16	Q4	1.5 est.	No	0.002
RTX 4070 Ti Super 16GB	16	Q4	1.4 est.	No	0.002
RX 7800 XT 16GB	16	Q4	1.3 est.	No	0.003
RX 9070 16GB	16	Q4	1.3 est.	No	0.002
RX 9070 XT 16GB	16	Q4	1.3 est.	No	0.002
RX 7900 GRE 16GB	16	Q4	1.2 est.	No	0.002
Arc A770 16GB	16	Q4	1.2 est.	No	0.003
Mac mini M4 32GB	32	Q4	1.0 est.	No	0.001
RTX 5060 Ti 16GB	16	Q4	0.9 est.	No	0.002
T4 16GB	16	Q4	0.7 est.	No	0.0
RTX 6000 Ada 48GB	48	FP16	0.6 est.	No	0.0
RTX 4060 Ti 16GB	16	Q4	0.6 est.	No	0.001
RX 7600 XT 16GB	16	Q4	0.6 est.	No	0.002
L40S 48GB	48	FP16	0.5 est.	No	0.0
RTX A6000 48GB	48	FP16	0.5 est.	No	0.0
Mac mini M4 Pro 64GB	64	FP16	0.3 est.	No	0.0

Fit checks

• RTX 5090 32GB — does not fit · ≈14.9 tok/s Q4 est.
• RTX 3090 Ti 24GB — does not fit · ≈4.7 tok/s Q4 est.
• RTX 4090 24GB — does not fit · ≈4.7 tok/s Q4 est.
• RTX 5080 16GB — does not fit · ≈2.0 tok/s Q4 est.
• RTX 6000 Ada 48GB — does not fit · ≈17.9 tok/s Q4 est.
• RX 7900 XTX 24GB — does not fit · ≈4.5 tok/s Q4 est.
• RTX 3090 24GB — does not fit · ≈4.4 tok/s Q4 est.
• RTX 5070 Ti 16GB — does not fit · ≈1.9 tok/s Q4 est.
• L40S 48GB — does not fit · ≈16.1 tok/s Q4 est.
• Mac Studio M3 Ultra 96GB — fits · ≈7.2 tok/s Q4 est.
• Mac Studio M2 Ultra 192GB — fits · ≈7.0 tok/s Q4 est.
• RX 7900 XT 20GB — does not fit · ≈2.6 tok/s Q4 est.

12 popular retail GPUs; every row in the table above also links its Fits? verdict to the full fit check.

Same family: Qwen

Qwen3 Qwen3-8B · Qwen3-14B
Qwen3 2507 Qwen3-4B-Instruct-2507 · Qwen3-4B-Thinking-2507 · Qwen3-30B-A3B-Instruct-2507 · Qwen3-235B-A22B-Instruct-2507
Qwen2.5 Qwen2.5-7B-Instruct · Qwen2.5-14B-Instruct · Qwen2.5-32B-Instruct
Qwen2.5 Coder Qwen2.5-Coder-32B-Instruct
Qwen3 Next Qwen3-Next-80B-A3B-Instruct
Qwen3 Coder Qwen3-Coder-480B-A35B-Instruct · Qwen3-Coder-Next
Qwen3.5 Qwen3.5-2B · Qwen3.5-9B · Qwen3.5-27B · Qwen3.5-35B-A3B · Qwen3.5-122B-A10B · Qwen3.5-397B-A17B
Qwen3.6 Qwen3.6-27B · Qwen3.6-35B-A3B

All Qwen models →