Qwen3 32B F16 — Hardware Requirements & GPU Compatibility
ChatReasoningFunctions
Qwen3 32B F16 is a 32B-parameter open language model from geoffmunn in the Qwen 3 family. At Q4_K_M it needs about 21.12 GB of VRAM — see which GPUs and Macs can run it below.
Decent on a 24 GB GPU
Runs on 24 GB (Q4_K_M · ~31 tok/s · 8K)Macs from 24 GB (Q2_K · ~14 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Qwen3 32B F16
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
vllm serve geoffmunn/Qwen3-32B-f16 --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 5090 (32 GB) (7 GB headroom) · Q4_K_M
- Generation speed
- ~55 tok/s
- generation speed
- Cost per 1M output tokens
- $2.50
- per 1M output tokens
How Much VRAM Does Qwen3 32B F16 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 15.0 GB | 23.1 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~44 tok/s | |
| 17.2 GB | 25.3 GB | 3-bit medium — 3-bit medium quantization | Fits: 16K context · ~38 tok/s | |
| 21.1 GB | 29.2 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 8K context · ~31 tok/s | |
| 25.1 GB | 33.2 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Does not fit: needs 28.9 GB | |
| 29.0 GB | 37.2 GB | 6-bit, very good — 6-bit quantization, very good quality | Does not fit: needs 32.8 GB | |
| 35.2 GB | 43.3 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 39 GB | |
| 70.4 GB | 78.5 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Does not fit: needs 74.2 GB |
Selected Q4_K_M
4-bit medium quantization24.9GBat 16K context
- Weights
- 20.3 GB
- KV cache
- 4.3 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · fits up to 8K context · ~31 tok/s
≥50 tok/s needs ≥1.62 TB/s · RTX 5090 ~55
Which Hardware Can Run Qwen3 32B F16?
Q4_K_M · 21.1 GBQwen3 32B F16 (Q4_K_M) requires 21.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 28+ GB is recommended. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 100 devices with unified memory can run Qwen3 32B F16, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M1 Max (32 GB).
20 GB
Q4_K_M · best fit Q2_K (18.7 GB)
1 GPU
RX 7900 XT~32Simulate output on RX 7900 XT32–40 GB
fits · 7.1 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~55Simulate output on RTX 5090M5 Max36 GB~15Simulate output on Mac Studio M5 Max (36 GB)M532 GB~5Simulate output on MacBook Air 13" M5 (32 GB)Orin~6Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~4Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Alternatives to Qwen3 32B F16
Qwen 3 family · 79 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Nanbeige4.3 33BNanbeige | — | 23.7 GB | ~32 | same size |
| Hy Memory 30B A3BTencent | — | 19.1 GB | ~184 | same size |
| Granite Switch 4.1 30BIBM | — | 22.7 GB | ~35 | same size |
| Llm Jp 4.1 32B A3b Thinkingllm-jp | — | 20.1 GB | ~188 | same size |
| Llm Jp 4.1 33B Thinkingllm-jp | — | 22.5 GB | ~35 | same size |
| Agnes 3.0 FlashAgnes-AI | — | 22.1 GB | ~35 | same size |
Specifications
- Publisher
- geoffmunn
- Family
- Qwen 3
- Parameters
- 32B
- Release Date
- 2025-10-12
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Qwen3 32B F16 need?
Qwen3 32B F16 requires 21.1 GB of VRAM at Q4_K_M, or 70.4 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 32B × 4.8 bits ÷ 8 = 19.2 GB
KV Cache + Overhead ≈ 1.9 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.
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run Qwen3 32B F16?
Yes, at Q4_K_M (21.1 GB) or lower. Higher quantizations like Q5_K_M (25.1 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for Qwen3 32B F16?
For Qwen3 32B F16, Q4_K_M (21.1 GB) offers the best balance of quality and VRAM usage. Q5_K_M (25.1 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 15.0 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Qwen3 32B F16 on a Mac?
Yes — Mac Mini M5 Pro (24 GB) and 106 other Macs can run Qwen3 32B F16, the smallest at Q2_K. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 21.1 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M1 Max (32 GB).
Can I run Qwen3 32B F16 locally?
Yes — Qwen3 32B F16 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Qwen3 32B F16?
At Q4_K_M, Qwen3 32B F16 can reach ~227 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~31 tok/s. 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 ÷ 21.1 × 0.65 = ~246 tok/s
Estimated speed at Q4_K_M (21.1 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Qwen3 32B F16?
At Q4_K_M, the download is about 19.20 GB. The full-precision BF16 version is 64.00 GB. The smallest option (Q2_K) is 13.60 GB.
Which GPUs can run Qwen3 32B F16?
7 consumer GPUs can run Qwen3 32B F16 at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU has plenty of headroom for comfortable inference.
Which devices can run Qwen3 32B F16?
100 devices with unified memory can run Qwen3 32B F16 at Q4_K_M (21.1 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (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.