Qwen1.5 72B Chat — Hardware Requirements & GPU Compatibility
Chat
Qwen1.5-72B-Chat is Alibaba's instruction-tuned, 72.3-billion-parameter flagship chat model from the Qwen1.5 series, a beta release of the Qwen2 architecture. Like the rest of the family, it offers stable 32K context length, broader multilingual support versus the original Qwen, and no need for custom trust_remote_code, and it was aligned on top of the pretrained base with supervised fine-tuning and direct preference optimization. As the largest dense model in the Qwen1.5 lineup, it was the series' primary competitor to other 70B-class open chat models at release. At 72.3 billion parameters, it needs a multi-GPU workstation or server to run, even once quantized. Context length is 32,768 tokens. It is released under Alibaba's Tongyi Qianwen license, a custom license that is free for most commercial and research use but requires a separate license from Alibaba once a deployment exceeds 100 million monthly active users. It was published in January 2024 and was later superseded by Qwen2-72B-Instruct.
Qwen1.5 72B Chat runs on a 64 GB+ Mac or a workstation GPU
Best local pick → Mac Pro M2 Ultra (64 GB) · Q4_K_M · 86.6 GB · ~11 tok/s
Macs from 64 GB (Q4_K_M · ~11 tok/s)
Other
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Qwen1.5 72B Chat
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
ollama run hf.co/mradermacher/Qwen1.5-72B-Chat-i1-GGUF:Q4_K_MQ4_K_M · needs ~86.6 GB · 16K context
Download: Qwen1.5-72B-Chat.i1-Q4_K_M.gguf44.10 GBmradermacher
Ways to runOther ways
each opens step-by-stepRun in cloud
Your own private copy · Q4_K_M · 96 GB GPU
- The easiest way to try this model
- Running in a few minutes, easy setup
- Chat right here or connect any app
$5+ bonus for new users
Runs on H200 (141 GB) (54 GB headroom) · Q4_K_M
- Generation speed
- ~64 tok/s
- generation speed
- Cost per 1M output tokens
- $15.68
- per 1M output tokens
How Much VRAM Does Qwen1.5 72B Chat Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 36.4 GB | 116.9 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Does not fit: needs 74 GB | |
| 37.3 GB | 117.8 GB | 3-bit small — 3-bit small quantization | Does not fit: needs 74.9 GB | |
| 40.9 GB | 121.4 GB | 3-bit medium — 3-bit medium quantization | Does not fit: needs 78.5 GB | |
| 41.8 GB | 122.3 GB | 4-bit · legacy — 4-bit legacy quantization | Does not fit: needs 79.4 GB | |
| 49.0 GB | 129.6 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Does not fit: needs 86.6 GB | |
| 57.2 GB | 137.7 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Does not fit: needs 94.8 GB | |
| 65.3 GB | 145.8 GB | 6-bit, very good — 6-bit quantization, very good quality | Does not fit: needs 102.9 GB | |
| 78.0 GB | 158.5 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 115.5 GB |
Selected Q4_K_M
4-bit medium quantization86.6GBat 16K context
- Weights
- 43.4 GB
- KV cache
- 43.0 GB
- Overhead
- 0.3 GB
Doesn't fit RTX 4090 (reference) · needs 62.6 GB more
Or .
Which Hardware Can Run Qwen1.5 72B Chat?
Q4_K_M · 49.0 GBQwen1.5 72B Chat (Q4_K_M) requires 49.0 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 64+ GB is recommended. Using the full 33K context window can add up to 80.5 GB, bringing total usage to 129.6 GB. No consumer GPU has enough memory. 52 devices with unified memory can run Qwen1.5 72B Chat, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (64 GB).
64–80 GB
tight · up to 4K ctx
3 Mac classes · 1 device family
M2 Ultra*64 GB~11Simulate output on Mac Pro M2 Ultra (64 GB)M5 Max64 GB~9Simulate output on Mac Studio M5 Max (64 GB)Orin~3Simulate output on NVIDIA Jetson AGX Orin 64GB96 GB
fits · 9.4 GB free · up to 16K ctx
2 Mac classes
M3 Ultra*96 GB~12Simulate output on Mac Studio (M3 Ultra, 96GB)M3 Max*96 GB~4Simulate output on MacBook Pro 14" M3 Max (96 GB)
By VRAM tier · typical GPUs and Macs at Q4_K_M
How good is Qwen1.5 72B Chat?
llmrun Score blends public benchmark boards into one number; 100 = the average open model.
Not enough data yet for an llmrun Score.
A Score needs results on enough independent boards. How the Composite Score is built
Benchmarks
Here are all 3 boards that have results for this model, sorted by their weight in the llmrun Score; badges show who ran the test.
- GPQA Diamond
Epoch AI · Feb 2024Independent
Independent28.8%
#85/100 open
- LMArena Text
LMArena · Oct 2026Arena
Arena1233.6
#135/188 open
- MMLU-Pro
TIGER-LabIndependent
Independent52.6%
#70/146 open
Alternatives to Qwen1.5 72B Chat
Qwen family · 28 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Llama 3.3 70B InstructMeta | 80.0 | 51.3 GB | —needs > 24 GB | same size |
| NVIDIA Nemotron 3.5 Bolt 74B A9B BF16NVIDIA | — | 54.2 GB | —needs > 24 GB | same size |
| CodeLlama 70B Instruct HFMeta | — | 23.2 GB | ~29 | same size |
| StableBeluga2Stability AI | — | 23.5 GB | ~29 | same size |
| Apertus 70B 2509swiss-ai | — | 48 GB | —needs > 24 GB | same size |
| DeepSeek R1 Distill Llama 70BDeepSeek | — | 48 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- Alibaba
- Family
- Qwen
- Parameters
- 72.3B
- Architecture
- Qwen2ForCausalLM
- Context Length
- 32,768 tokens
- Vocabulary Size
- 152,064
- Release Date
- 2024-01-30
- License
- Other
Frequently Asked Questions
How much VRAM does Qwen1.5 72B Chat need?
Qwen1.5 72B Chat requires 49.0 GB of VRAM at Q4_K_M, or 150.2 GB at BF16. Full 33K context adds up to 80.5 GB (129.6 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 72.3B × 4.8 bits ÷ 8 = 43.4 GB
KV Cache + Overhead ≈ 5.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 86.2 GB (at full 33K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Qwen1.5 72B Chat?
Yes, at IQ2_M (30.1 GB) or lower. Higher quantizations like IQ3_XXS (33.7 GB) exceed the NVIDIA GeForce RTX 5090's 32 GB.
What's the best quantization for Qwen1.5 72B Chat?
For Qwen1.5 72B Chat, Q4_K_M (49.0 GB) offers the best balance of quality and VRAM usage. Q5_K_S (55.4 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 25.6 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Qwen1.5 72B Chat on a Mac?
Yes — Mac Pro M2 Ultra (64 GB) and 41 other Macs can run Qwen1.5 72B Chat. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 49.0 GB of usable unified memory (RAM minus macOS overhead).
Can I run Qwen1.5 72B Chat locally?
Yes — Qwen1.5 72B Chat can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Qwen1.5 72B Chat?
At Q4_K_M, Qwen1.5 72B Chat can reach ~98 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 49.0 × 0.65 = ~106 tok/s
Estimated speed at Q4_K_M (49.0 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Qwen1.5 72B Chat?
At Q4_K_M, the download is about 43.37 GB. The full-precision BF16 version is 144.58 GB. The smallest option (IQ2_XXS) is 19.88 GB.
Which GPUs can run Qwen1.5 72B Chat?
No single consumer GPU has enough VRAM to run Qwen1.5 72B Chat at Q4_K_M (49.0 GB). Multi-GPU or professional hardware is required.
Which devices can run Qwen1.5 72B Chat?
52 devices with unified memory can run Qwen1.5 72B Chat at Q4_K_M (49.0 GB), including ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB), GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.