What GPU do I need to run Qwen/Qwen3-235B-A22B-Thinking-2507?

What Qwen3-235B-A22B-Thinking-2507 is

Qwen3-235B-A22B-Thinking-2507 is a 235B-parameter mixture-of-experts language model published by Alibaba (Qwen) on Hugging Face. It works through a problem step by step before answering, rather than responding directly. It is released under Apache 2.0.

License note: fully permissive, no gating and no usage restrictions. Facts in this section are sourced from Qwen3-235B-A22B-Thinking-2507's Hugging Face model card, not benchmarked by Aquanode.

What it's used for

• Multi-step reasoning and math
• Code generation
• Agentic tool use

VRAM required & cheapest live GPU fit

Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.

A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.

INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.

Qwen3-235B-A22B-Thinking-2507: KV cache by context length

The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.

Grouped-query attention: every layer caches keys and values for a few shared KV heads. Cached per token: 192,512 bytes at 16-bit.

Computed from the layer, head and window counts in the model's own configuration (Qwen/Qwen3-235B-A22B-Thinking-2507), at 16 bits per cached value. Each concurrent sequence needs its own cache, on top of the weights.

Qwen3-235B-A22B-Thinking-2507: common questions

How to run Qwen3-235B-A22B-Thinking-2507

vllm serve Qwen/Qwen3-235B-A22B-Thinking-2507 --tensor-parallel-size 8 --max-model-len 262144 --enable-reasoning --reasoning-parser deepseek_r1

llama-server -hf unsloth/Qwen3-235B-A22B-Thinking-2507-GGUF

Deploy Qwen3-235B-A22B-Thinking-2507 on Aquanode

Aquanode has no one-click deploy template for Qwen3-235B-A22B-Thinking-2507; you install the inference engine yourself with the commands below. Aquanode sells GPU pods billed per second, not a hosted inference API.

1. Launch a bare GPU pod sized to the requirement above (3× AMD MI300X or larger).
2. Open a terminal on the pod, or save one of the commands above as a startup script so it runs automatically the first time the pod boots.
3. Run the command and connect to the resulting endpoint.

Launch a GPU pod

More Qwen3 models

• Qwen3-235B-A22B (235.1B, BF16)
• Qwen3-235B-A22B-Instruct-2507 (235.1B, BF16)
• Qwen3-235B-A22B-Instruct-2507-FP8 (235.1B, F8_E4M3)
• Qwen3-235B-A22B-Thinking-2507-FP8 (235.1B, F8_E4M3)
• Qwen3-235B-A22B-FP8 (235.1B, F8_E4M3)

All 100 Qwen3 models: VRAM and GPU requirements

Lineage

Quantized and fine-tuned versions

• Qwen3-235B-A22B-Thinking-2507-FP8

Alternatives at this size

Other models for text-generation within about a third of Qwen3-235B-A22B-Thinking-2507's 235.1B parameters, from other model lines.

• MiniMax-M2.7 (228.7B, F8_E4M3)
• MiMo-V2.5 (310.8B, F8_E4M3)
• Step-3.5-Flash (199.4B, BF16)
• Hy3-preview (298.8B, BF16)
• DeepSeek-V2 (235.7B, BF16)

More on Qwen3-235B-A22B-Thinking-2507

• Fits on a 288 GB GPU at FP8: every model that fits in 288 GB.

• Fits on a 141 GB GPU at INT4: every model that fits in 141 GB.

• Best reasoning models: how Qwen3-235B-A22B-Thinking-2507 ranks against the rest.

• Best chat and assistants models: how Qwen3-235B-A22B-Thinking-2507 ranks against the rest.

• How to run Qwen3 and how to fine-tune Qwen3.