Can I run it?
Can I run Qwen3.8 35B A3B Distill on GeForce RTX 3090?
Tight fit
Tight fit fully on your GeForce RTX 3090 at Q4_K_M, 8K context, about 257–351 tok/s. Weights 20.2 GB plus 0.19 GB of KV cache — 20.9 GB of 22.3 GB usable. Up to 32K context still fits.
Best quantQ4_K_M
Working context8K
Max that fits32K
Speed257–351 tok/s
Weights20.2 GB
KV cache0.19 GB
Total20.9 GB
Usable22.3 GB
Qwen3.8 35B A3B Distill details · GeForce RTX 3090 page · How this estimate is made
Run commands
llama.cpp
llama-server -hf empero-ai/Qwen3.8-35B-A3B-Distill-GGUF:Q4_K_M -c 8192 -ngl 40
Ollama
ollama run qwen3.8
LM Studio
lms get empero-ai/Qwen3.8-35B-A3B-Distill-GGUF # Then load quant Q4_K_M
Assumptions
- Estimate, not a benchmark. Decode speed uses a memory-bandwidth model at 55–75% of peak.
- KV cache is FP16. Engine overhead assumes llama.cpp (0.3 GB + 1% of the weights). Ollama is closer to 0.9 GB + 3%, LM Studio closer to 1 GB + 3%.
- Q4_K_M size is the real GGUF file size from Hugging Face.
- 0.5 GB is reserved because this GPU also drives a display, then 5% of the rest is left free.
- Speed is quoted at 8K context, not at the maximum context that still fits.
Larger models that also fit
- Can I run Cyber Tiel Coder 35B A3B GGUF MTP on GeForce RTX 3090? — Tight fit, Q4_K_M, 259–353 tok/s
- Can I run DeepSeek R1 Distill Llama 70B on GeForce RTX 3090? — Tight fit, UD-IQ2_XXS, 23–32 tok/s
- Can I run Llama 3.3 70B Instruct on GeForce RTX 3090? — Tight fit, UD-IQ2_XXS, 23–32 tok/s
Updated 2026-10-24. Profile on this page is the device default plus 32 GB of DDR5-5600 system RAM.