Loading…
Loading…
Models / Qwen3-Next
2 models (80B to 80B) · ↓ 439.8K downloads a month · from $0.78/h on 4x RTX 4060 Ti right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-Next-80B-A3B-Instruct | vLLM | 163 GB | 183.8 GB | $2.36/h 4x Nvidia A40 |
| FP8Qwen/Qwen3-Next-80B-A3B-Instruct-FP8 | vLLM | 82 GB | 94.3 GB | $1.38/h 8x RTX 3090 |
| NVFP4 (nvidia)nvidia/Qwen3-Next-80B-A3B-Instruct-NVFP4 | vLLM | 51 GB | 59.5 GB | $0.89/h 2x Nvidia A40 |
| bnb 4bit (unsloth)unsloth/Qwen3-Next-80B-A3B-Instruct-bnb-4bit | vLLM | 42 GB | 50 GB | $0.78/h 4x RTX 4060 Ti |
| AWQ 4bit (cyankiwi)cyankiwi/Qwen3-Next-80B-A3B-Instruct-AWQ-4bit | vLLM | 49 GB | 57.8 GB | $0.89/h 2x Nvidia A40 |
| FP8 (RedHatAI)RedHatAI/Qwen3-Next-80B-A3B-Instruct-FP8 | vLLM | 82 GB | 94.3 GB | $1.38/h 8x RTX 3090 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3-Next-80B-A3B-Thinking | vLLM | 163 GB | 183.8 GB | $2.36/h 4x Nvidia A40 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").