Can Arc Pro B70 32GB run NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4?

Intel · 32 GB GDDR6 · 608 GB/s bandwidth · NVIDIA · 550.0B (active 55.0B) · MoE · model ctx up to 256K

ONEAPI

No — NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 at Q4 needs ≈ 303.77 GB even at 4K context, beyond the Arc Pro B70 32GB's 32 GB GDDR6. It would only run with heavy CPU/disk offload. FP16 needs ≈ 1101.55 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	303.77 GB	No	—
Q4	8K default	304.05 GB	No	—
Q4	32K	305.7 GB	No	—
FP16	4K	1101.28 GB	No	—
FP16	8K default	1101.55 GB	No	—
FP16	32K	1103.2 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	Arc Pro B70 32GB VRAM
Q4	302.5 GB	0.55 GB	1 GB	304.05 GB	32 GB
FP16	1100.0 GB	0.55 GB	1 GB	1101.55 GB	32 GB

Other GPUs that run NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4

• DGX Spark (GH200 480GB) — ≈5.8 tok/s Q4 est.

All GPUs for NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 →

Other models for the Arc Pro B70 32GB

• MiniCPM5-1B — ≈358.2 tok/s Q4 est.
• Qwen3.5-2B — ≈193.5 tok/s Q4 est.
• DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx — ≈168.2 tok/s Q4 est.
• Qwen3.6-35B-A3B — ≈129.0 tok/s Q4 est.
• Qwen3.5-35B-A3B — ≈129.0 tok/s Q4 est.
• GLM-4.7-Flash — ≈129.0 tok/s Q4 est.
• Qwen3-30B-A3B-Instruct-2507 — ≈129.0 tok/s Q4 est.
• Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 — ≈129.0 tok/s Q4 est.

All models on the Arc Pro B70 32GB →