Can Titan V 12GB run Qwen3.5-122B-A10B?

NVIDIA · 12 GB HBM2 · 652.8 GB/s bandwidth · Alibaba · 122.0B (active 10.0B) · MoE · model ctx up to 256K

CUDA

No — Qwen3.5-122B-A10B at Q4 needs ≈ 68.23 GB even at 4K context, beyond the Titan V 12GB's 12 GB HBM2. It would only run with heavy CPU/disk offload. FP16 needs ≈ 245.25 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	68.23 GB	No	—
Q4	8K default	68.35 GB	No	—
Q4	32K	69.1 GB	No	—
FP16	4K	245.12 GB	No	—
FP16	8K default	245.25 GB	No	—
FP16	32K	246.0 GB	No	—

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).

VRAM breakdown at 8K context

Quant	Weights	KV cache	Overhead	Total needed	Titan V 12GB VRAM
Q4	67.1 GB	0.25 GB	1 GB	68.35 GB	12 GB
FP16	244.0 GB	0.25 GB	1 GB	245.25 GB	12 GB

Other GPUs that run Qwen3.5-122B-A10B

• B200 192GB SXM — ≈509.1 tok/s Q4 est.
• MI300X 192GB — ≈337.3 tok/s Q4 est.
• H200 141GB SXM5 — ≈305.5 tok/s Q4 est.
• H100 80GB SXM5 — ≈213.2 tok/s Q4 est.
• Gaudi 3 128GB — ≈208.5 tok/s Q4 est.
• A100 80GB SXM4 — ≈129.8 tok/s Q4 est.
• Mac Studio M3 Ultra 96GB — ≈52.1 tok/s Q4 est.
• Mac Studio M2 Ultra 192GB — ≈50.9 tok/s Q4 est.

All GPUs for Qwen3.5-122B-A10B →

Other models for the Titan V 12GB

• MiniCPM5-1B — ≈384.6 tok/s Q4 est.
• Qwen3.5-2B — ≈207.7 tok/s Q4 est.
• DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx — ≈180.6 tok/s Q4 est.
• Llama-3.2-3B-Instruct — ≈129.8 tok/s Q4 est.
• Qwen3-4B-Instruct-2507 — ≈103.9 tok/s Q4 est.
• NVIDIA-Nemotron-3-Nano-4B-BF16 — ≈103.9 tok/s Q4 est.
• Qwen3-4B-Thinking-2507 — ≈103.9 tok/s Q4 est.
• gemma-4-E2B-it — ≈81.5 tok/s Q4 est.

All models on the Titan V 12GB →