Qwen3.5-397B-A17B
Qwen- 403.4B parameters
- Mixture of experts, 17B active
- Apache-2.0
- Released in bf16
- Estimated
Qwen/Qwen3.5-397B-A17B on Hugging Face
Qwen3.5-397B-A17B with Glyd
| bf16 | Glyd | Change, % | ||
|---|---|---|---|---|
| Weights in GPU memory, GBEstimated from the parameter count | 806.80 | 542.97est. | −33 | |
| 8×H100 nodes it takesWeights and an 8K-token KV cache | 2 | 1 | ||
| H100 80 GB GPUs it takes | 10 | 7 |
Which single GPU it fits
Estimated from the parameter count.
| Memory | GPUs | bf16 | Glyd |
|---|---|---|---|
| 16 GB | RTX 4080, RTX 5080 | No | No |
| 24 GB | RTX 4090, RTX 3090, A10 | No | No |
| 32 GB | RTX 5090 | No | No |
| 48 GB | RTX A6000, L40S, RTX 6000 Ada | No | No |
| 80 GB | H100, A100 80 GB | No | No |
| 96 GB | RTX PRO 6000, GH200 | No | No |
| 141 GB | H200 | No | No |
Related models
Measured so far
Twenty-two open models measured, with quality wherever bf16 fits one GPU.