Can an NVIDIA GeForce RTX 3090 run Qwen3.6-35B-A3B?
Just barely. Qwen3.6-35B-A3B needs 22.8GB at MXFP4_MOE with llama.cpp and 8K context; the NVIDIA GeForce RTX 3090 has 24GB of VRAM (23.5GB usable) — grade C: it loads, but there is little headroom for longer context.
NVIDIA GeForce RTX 3090 · Qwen3.6-35B-A3B · MXFP4_MOE
⚠️ Tight — grade C
InteractivebetaHybrid attention — KV computed as full-cache upper bound · MoE speed estimate — wide error bars
weights 21.7 GBKV cache 0.67 GBoverhead 0.42 GBpool 23.5 GBneeds 22.8 GB
assumes llama.cpp · batch 1 · 8K context · fp16 KV cache
$1,499 MSRP · as of 2026-07-08
Affiliate link — we may earn a commission. Verdicts are computed before any link is attached.
Every quant, graded
NVIDIA GeForce RTX 3090 × Qwen3.6-35B-A3B, llama.cpp, 8K context.
| Quant | File size | Fit | Speed tier |
|---|---|---|---|
| Q8_0 | 36.9 GB | Tight · C | interactive · offload · beta |
| UD-Q5_K_XL | 26.6 GB | Tight · C | interactive · offload · beta |
| UD-Q4_K_XL | 22.4 GB | Tight · C | interactive · beta |
| MXFP4_MOE | 21.7 GB | Tight · C | interactive · beta |
Qwen3.6-35B-A3B on similar hardware
More models on the NVIDIA GeForce RTX 3090
Verdict computed from published specs and measured GGUF file sizes — see how the math works.