Qwen3 Coder 480B A35B GPU requirements: 240 GB at 4-bit, 960 GB at BF16
Qwen3 Coder 480B A35B needs 960 GB of GPU memory for its weights at BF16, 480 GB at FP8 and 240 GB at 4-bit, plus 8.32 GB of KV cache for a 32K context. The cheapest node that runs it here is 4× A100 PCIe 80GB at 4-bit, for $0.36 an hour on spot.
Weights plus a 32K 16-bit KV cache, with 10% of the node's memory kept free. Billed per minute.
- Parameters
- 480B35B active per token
- Weights, 4-bit
- 240 GBBF16: 960 GB
- KV cache, 32K
- 8.32 GBOne sequence, 16-bit cache
- Context window
- 256KNative
How much VRAM Qwen3 Coder 480B A35B needs, format by format.
Weights are parameters × bytes per parameter. Each context column adds the KV cache of one sequence of that length.
| Format | Weights | + 8Kcontext | + 32Kcontext | + 128Kcontext | + 256Kcontext | Cheapest node, 32K |
|---|---|---|---|---|---|---|
| BF16Published weights, fine-tuning | 960 GB | 962 GB | 969 GB | 994 GB | 1.03 TB | 8× H200 NVL · $5.52/h |
| FP8Hopper, Ada, Blackwell, MI300 | 480 GB | 482 GB | 488 GB | 513 GB | 547 GB | 8× RTX PRO 6000 Blackwell · $1.52/h |
| 4-bitAWQ, GPTQ, GGUF Q4 | 240 GB | 242 GB | 248 GB | 273 GB | 307 GB | 4× A100 PCIe 80GB · $0.36/h |
The serving engine adds its own buffers, and an 8-bit KV cache halves the context columns. We keep 10% of a node's memory free when we say a model fits. Speed is a separate question: it follows memory bandwidth and, for a mixture of experts, the active parameters.
GPUs that run Qwen3 Coder 480B A35B, cheapest first.
For each format, the smallest node of each GPU model that holds the weights and a 32K context, sorted by spot price. One row per GPU model; larger nodes of the same GPU work too.
| Format | GPU | GPU memory | Spot | On-demand | 730 h spot | Action |
|---|---|---|---|---|---|---|
| 4-bit | 4× A100 PCIe 80GBAmpere · 248 GB needed | 320 GB | $0.36 | $0.76 | $262.80 | Deploy |
| 4-bit | 8× A40Ampere · 248 GB needed | 384 GB | $0.48 | $0.72 | $350.40 | Deploy |
| 4-bit | 8× A100 40GBAmpere · 248 GB needed | 320 GB | $0.72 | $1.36 | $525.60 | Deploy |
| 4-bit | 8× L40Ada Lovelace · 248 GB needed | 384 GB | $0.72 | $1.52 | $525.60 | Deploy |
| FP8 | 8× RTX PRO 6000 BlackwellBlackwell · 488 GB needed | 768 GB | $1.52 | $3.12 | $1,109.60 | Deploy |
| FP8 | 4× H200 NVLHopper · 488 GB needed | 564 GB | $2.76 | $3.96 | $2,014.80 | Deploy |
| FP8 | 4× H200 SXMHopper · 488 GB needed | 564 GB | $3.00 | $4.36 | $2,190.00 | Deploy |
| FP8 | 8× H100 PCIeHopper · 488 GB needed | 640 GB | $3.60 | $5.52 | $2,628.00 | Deploy |
| BF16 | 8× H200 NVLHopper · 969 GB needed | 1128 GB | $5.52 | $7.92 | $4,029.60 | Deploy |
| BF16 | 8× H200 SXMHopper · 969 GB needed | 1128 GB | $6.00 | $8.72 | $4,380.00 | Deploy |
| BF16 | 8× MI300XCDNA 3 · 969 GB needed | 1536 GB | $7.92 | $11.60 | $5,781.60 | Deploy |
| BF16 | 8× MI325XCDNA 3 · 969 GB needed | 2048 GB | $8.72 | $12.72 | $6,365.60 | Deploy |
Spot capacity can be reclaimed with a 2-minute notice; for a live endpoint, use on-demand. How spot works · serving a language model.
What a long context costs in memory.
Grouped-query attention, 8 KV heads of 128 across 62 layers.
| Context | KV cache, 1 sequence | 8 sequences |
|---|---|---|
| 8K tokens | 2.08 GB | 16.6 GB |
| 32K tokens | 8.32 GB | 66.6 GB |
| 128K tokens | 33.3 GB | 266 GB |
| 256K tokens | 66.6 GB | 533 GB |
Serving several users
KV cacheEvery concurrent request keeps its own cache, so memory for context grows with users × context length. A server batching 8 requests of 32K needs eight times the single-sequence figure — 66.6 GB here.
- Recommended node4× A100 PCIe 80GB
- Weights at 4-bit240 GB
- Context that fits, 1 sequence184K
Paged attention (vLLM, SGLang) allocates the cache in blocks as requests grow, so a node holds more short requests than the worst case suggests.
Qwen3 Coder 480B A35B, as published.
- Developer
- Alibaba QwenAgentic coding model, repository-scale context
- Parameters
- 480B total, 35B active per token
- Architecture
- Mixture of experts, 62 layersGrouped-query attention, 8 KV heads of 128 across 62 layers
- Context window
- 262,144 tokensAs set in the published configuration
- Published format
- BF16, 960 GB
- Licence
- Apache 2.0
- Released
- 22 July 2025
- Source
- Qwen/Qwen3-Coder-480B-A35B-Instructconfig.json read 26 October 2026
Serve it with vLLM
template: vLLMOn 4× A100 PCIe 80GB, from the instance terminal:
$ vllm serve Qwen/Qwen3-Coder-480B-A35B-Instruct \ --tensor-parallel-size 4 \ --max-model-len 32768
Very recent architectures need a recent vLLM; the model card gives the minimum version and any extra flags.
Qwen3 Coder 480B A35B on a GPU — the questions that come up
How much VRAM does Qwen3 Coder 480B A35B need?
For the weights alone: 960 GB at BF16, 480 GB at FP8 and 240 GB at 4-bit. Add 8.32 GB of KV cache for one 32K-token context, or 66.6 GB at the full 256K window, and keep about 10% of the memory free for activations.
Can Qwen3 Coder 480B A35B run on a single GPU?
One RTX 4090 (24 GB): no. One RTX 5090 (32 GB): no. One H100 SXM (80 GB): no. One H200 SXM (141 GB): no. One B200 SXM (180 GB): no. Those answers assume a 32K context and the widest format that still fits.
What is the cheapest way to run Qwen3 Coder 480B A35B?
Here, 4× A100 PCIe 80GB at 4-bit: $0.36 per hour on spot and $0.76 on-demand, billed per minute — $262.80 for a 730-hour month on spot. That node has 320 GB of GPU memory for the 248 GB this setup needs.
Does Qwen3 Coder 480B A35B need less memory because only 35B parameters are active?
No. A mixture-of-experts model routes each token through a few experts, but every expert has to be resident in GPU memory, so the memory bill follows the 480B total. What the 35B active parameters change is the work per token: generation reads far less of the weights each step than a dense model of the same size would, which is why it is fast for its size.
Which format should I run it in?
BF16 keeps the published weights untouched and is the format for fine-tuning. FP8 halves the memory and runs in hardware on Hopper, Ada, Blackwell and MI300-class GPUs. 4-bit quantisation (AWQ, GPTQ, GGUF) quarters it, at a quality cost that is larger on small models than on big ones; check it against your own evaluation before relying on it.
How do I deploy it?
Pick the node in the table above and launch it with the vLLM template; the model downloads from Hugging Face (Qwen/Qwen3-Coder-480B-A35B-Instruct) on first start. Spread the weights over the node with --tensor-parallel-size equal to the GPU count, and set --max-model-len to the context you need. New architectures need a recent vLLM: check the model card for the minimum version.
Sizing another model? GPU requirements for 64 open models · serving a language model · fine-tuning.
Qwen3 Coder 480B A35B on 4× A100 PCIe 80GB for $0.36 an hour.
Launch the node with the vLLM template, point it at Qwen/Qwen3-Coder-480B-A35B-Instruct, stop it when you are done. Prepaid, billed by the minute.
Billed per minute from the moment the instance is reachable. Minimum credit $30, no subscription.