NVIDIA · 12 GB GDDR6X · 554 GB/s bandwidth ·
Alibaba · 8.95B · model ctx up to 256K
Yes — the RTX 4070 Ti 12GB (12 GB GDDR6X) can run Qwen3.5-9B. At Q4 with the default 8K context it needs ≈ 7.71 GB VRAM (weights 4.92 GB + KV cache 1.79 GB + 1 GB runtime overhead), which fits within 12 GB. Decode at Q4/8K: ≈ 39.4 tok/s (estimated, batch 1). FP16 needs ≈ 20.69 GB — does not fit on this card. Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.
Fit & speed by quant and context
Quant
Context
VRAM needed
Fits?
tok/s (decode)
Q4
4K
6.82 GB
Yes
39.4est.
Q4
8K default
7.71 GB
Yes
39.4est.
Q4
32K
13.08 GB
No
94.7est.
FP16
4K
19.79 GB
No
11.4est.
FP16
8K default
20.69 GB
No
10.4est.
FP16
32K
26.06 GB
No
6.6est.
Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured
anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s
means the model is far beyond this card (offload-only territory).