# Qwen3 235B-A22B (2507): hardware requirements
> Qwen3 235B-A22B (2507) (Alibaba Qwen) is a 235B (22B active) mixture-of-experts model with a 256K-token context window, released under the Apache 2.0 license. It needs about 149GB of memory at Q4_K_M with 8K context.
Qwen3 235B-A22B was Alibaba's open flagship for 2025, and its July 2025 (2507) Instruct and Thinking refreshes were among the best open models of that year. It has been superseded by the Qwen3.5 generation: Qwen3.5 122B-A10B matches or beats it in about half the memory, and Qwen3.5 397B-A17B replaced it at the top. At Q4_K_M it needs roughly 150GB with 8K context, which means a 192GB or larger Mac Studio, a multi-GPU workstation, or a server with lots of RAM for expert offload. With 22B active parameters it is noticeably slower per token than newer 10B to 17B active designs.

## Recommended hardware
- **Multi-GPU:** 8× RTX 3090 (192GB), Q4_K_M, ~18 tokens/sec
- **Multi-GPU:** 8× RTX 4090 (192GB), Q4_K_M, ~20 tokens/sec
- **Multi-GPU:** 8× RTX 5090 (256GB), Q4_K_M, ~27 tokens/sec
- **Mac / unified memory:** Ryzen AI Max+ PRO 495 (192GB), Q3_K_M, ~11 tokens/sec

## Memory by quantization
| Quant | Weights | Total @ 8K ctx | Total @ 32K ctx |
| --- | --- | --- | --- |
| Q2_K | 88.1GB | 92.9GB | 97.7GB |
| Q3_K_M | 114.6GB | 120.2GB | 124.9GB |
| Q4_K_M | 142.5GB | 148.9GB | 153.7GB |
| Q5_K_M | 167.4GB | 174.6GB | 179.4GB |
| Q6_K | 193.9GB | 201.9GB | 206.6GB |
| Q8_0 | 249.7GB | 259.4GB | 264.1GB |
| BF16 | 470.0GB | 486.3GB | 491.0GB |

## GPUs that run it fully in memory (fastest first)
| GPU | Memory | Quant | Speed | Price |
| --- | --- | --- | --- | --- |
| M5 Ultra 256GB | 256GB | Q4_K_M | ~27 tok/s | ~$9,499 |
| M5 Ultra 512GB | 512GB | Q8_0 | ~16 tok/s | n/a |
| M3 Ultra 512GB | 512GB | Q8_0 | ~11 tok/s | $9,499 MSRP |
| Ryzen AI Max+ PRO 495 | 192GB | Q3_K_M | ~11 tok/s | ~$6,800 |

## How to run it
- Ollama: `ollama run qwen3:235b`
- llama.cpp: `llama-server -hf unsloth/Qwen3-235B-A22B-Instruct-2507-GGUF:Q4_K_M -c 8192 -ngl 99`
- Weights: https://huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507

## Strengths
- Broad knowledge and strong writing for an open model of its generation
- 262K native context and separate Instruct and Thinking checkpoints
- Apache 2.0 license

## Weaknesses
- Needs about 150GB at Q4_K_M, beyond any single consumer GPU
- Superseded by Qwen3.5 122B-A10B, which is about as capable in half the memory
- 22B active parameters make it slower than newer sparse models on the same hardware
- Text only
Full page: https://llamabuilds.ai/models/qwen3-235b-a22b · Hardware guide: https://llamabuilds.ai/models/qwen3-235b-a22b/requirements

---
Source: https://llamabuilds.ai/models/qwen3-235b-a22b · Updated October 14, 2026 · Llama Builds. Speeds marked ~ are estimates from our bandwidth model (https://llamabuilds.ai/methodology); measured results are labeled as such.