Loading…
Loading…
Models / Qwen3.6
2 models (27B to 35B) · ↓ 22.2M downloads a month · from $0.34/h on 2x RTX 5060 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/Qwen3.6-27B | vLLM | 56 GB | 66.3 GB | $0.89/h 2x Nvidia A40 |
| FP8Qwen/Qwen3.6-27B-FP8 | vLLM | 31 GB | 38.9 GB | $0.45/h Nvidia A40 |
| NVFP4 (nvidia)nvidia/Qwen3.6-27B-NVFP4 | vLLM | 22 GB | 29 GB | $0.34/h 2x RTX 5060 Ti |
| NVFP4 (unsloth)unsloth/Qwen3.6-27B-NVFP4 | vLLM | 23 GB | 30.6 GB | $0.41/h RTX PRO 4500 |
| AWQ INT4 (cyankiwi)cyankiwi/Qwen3.6-27B-AWQ-INT4 | vLLM | 20 GB | 27.3 GB | $0.34/h 2x RTX 5060 Ti |
| int4 AutoRound (Lorbus)Lorbus/Qwen3.6-27B-int4-AutoRound | vLLM | 19 GB | 25.7 GB | $0.34/h 2x RTX 5060 Ti |
| AWQ 4bit (Alennnndy)Alennnndy/Qwen3.6-27B-AWQ-4bit | vLLM | 19 GB | 25.3 GB | $0.34/h 2x RTX 5060 Ti |
| GPTQ 4bit (btbtyler09)btbtyler09/Qwen3.6-27B-GPTQ-4bit | vLLM | 22 GB | 28.9 GB | $0.34/h 2x RTX 5060 Ti |
| AWQ (QuantTrio)QuantTrio/Qwen3.6-27B-AWQ | vLLM | 22 GB | 28.9 GB | $0.34/h 2x RTX 5060 Ti |
| Text NVFP4 MTP (sakamakismile)sakamakismile/Qwen3.6-27B-Text-NVFP4-MTP | vLLM | 20 GB | 26.4 GB | $0.34/h 2x RTX 5060 Ti |
| NVFP4 (mmangkad)mmangkad/Qwen3.6-27B-NVFP4 | vLLM | 31 GB | 39 GB | $0.45/h Nvidia A40 |
| NVFP4 GPTQ (sudo-0x2a)sudo-0x2a/Qwen3.6-27B-NVFP4-GPTQ | vLLM | 21 GB | 27.5 GB | $0.34/h 2x RTX 5060 Ti |
| OBLITERATED (OBLITERATUS)OBLITERATUS/Qwen3.6-27B-OBLITERATED | vLLM | 54 GB | 64.4 GB | $0.89/h 2x Nvidia A40 |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalQwen/Qwen3.6-35B-A3B | vLLM | 72 GB | 82.9 GB | $0.89/h 2x Nvidia A40 |
| FP8Qwen/Qwen3.6-35B-A3B-FP8 | vLLM | 37 GB | 44.6 GB | $0.45/h Nvidia A40 |
| NVFP4 (nvidia)nvidia/Qwen3.6-35B-A3B-NVFP4 | vLLM | 23 GB | 29 GB | $0.34/h 2x RTX 5060 Ti |
| NVFP4 Fast (unsloth)unsloth/Qwen3.6-35B-A3B-NVFP4-Fast | vLLM | 24 GB | 29.2 GB | $0.34/h 2x RTX 5060 Ti |
| NVFP4 (RedHatAI)RedHatAI/Qwen3.6-35B-A3B-NVFP4 | vLLM | 25 GB | 30.8 GB | $0.41/h RTX PRO 4500 |
| AWQ (QuantTrio)QuantTrio/Qwen3.6-35B-A3B-AWQ | vLLM | 25 GB | 31.3 GB | $0.41/h RTX PRO 4500 |
| NVFP4 (unsloth) |
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").
| vLLM |
| 26 GB |
| 32.4 GB |
| $0.45/h Nvidia A40 |
| AWQ 4bit (cyankiwi)cyankiwi/Qwen3.6-35B-A3B-AWQ-4bit | vLLM | 25 GB | 30.7 GB | $0.41/h RTX PRO 4500 |
| GPTQ Int4 (palmfuture)palmfuture/Qwen3.6-35B-A3B-GPTQ-Int4 | vLLM | 24 GB | 30.1 GB | $0.41/h RTX PRO 4500 |
| INT8 (lokeshe09)lokeshe09/Qwen3.6-35B-A3B-INT8 | vLLM | 36 GB | 42.7 GB | $0.45/h Nvidia A40 |
| PrismaQuant 4.75bit vllm (rdtand)rdtand/Qwen3.6-35B-A3B-PrismaQuant-4.75bit-vllm | vLLM | 23 GB | 28.4 GB | $0.34/h 2x RTX 5060 Ti |
| PRISM NVFP4 (Ex0bit)Ex0bit/Qwen3.6-35B-A3B-PRISM-NVFP4 | vLLM | 25 GB | 30.8 GB | $0.41/h RTX PRO 4500 |
| Uncensored NVFP4 (kyaky)kyaky/Qwen3.6-35B-A3B-Uncensored-NVFP4 | vLLM | 23 GB | 28 GB | $0.34/h 2x RTX 5060 Ti |