NVIDIA-Nemotron-3-Ultra-550B-A55B GGUF size and VRAM requirements

License: other ⬇ 13,541 ❤ 33
Parameters549.31B
Context1,048,576

unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF is a very large language model with 549.31 billion parameters, built on the nemotron_h_moe architecture. It is released under the other license and has been downloaded 13,541 times.

To run unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF locally at a 4,096-token context, its quantized versions need between 187.23 GB (IQ1_M, lowest quality) and 1035.67 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.

→ Guide: How much VRAM do you need?

GGUF file size and memory by quantization

Compare real GGUF weight sizes, estimated KV cache and total memory for Q4, Q5, Q8 and every quantization published in this repository.

Quant.Bits QualityWeights KVTotal Speed~Verdict
IQ1_M 2.74 Low 175.12 GB 11.31 GB 187.23 GB Insufficient
IQ2_XXS 2.82 Low 180.51 GB 11.31 GB 192.62 GB Insufficient
IQ2_M 2.82 Low 180.64 GB 11.31 GB 192.75 GB Insufficient
Q2_K_XL 2.94 Low 188.3 GB 11.31 GB 200.41 GB Insufficient
IQ3_XXS 3.25 Low 207.99 GB 11.31 GB 220.1 GB Insufficient
IQ3_S 3.65 Fair 233.3 GB 11.31 GB 245.41 GB Insufficient
Q3_K_M 3.99 Fair 255.21 GB 11.31 GB 267.32 GB Insufficient
Q3_K_XL 3.99 Fair 255.45 GB 11.31 GB 267.56 GB Insufficient
IQ4_XS 4.17 Fair 266.47 GB 11.31 GB 278.58 GB Insufficient
IQ4_NL 4.28 Good 273.97 GB 11.31 GB 286.08 GB Insufficient
Q4_K_S 4.77 Good 305.18 GB 11.31 GB 317.29 GB Insufficient
GGUF 5.13 Very good 328.08 GB 11.31 GB 340.19 GB Insufficient
Q4_K_M 5.23 Very good 334.55 GB 11.31 GB 346.66 GB Insufficient
Q4_K_XL 5.25 Very good 335.42 GB 11.31 GB 347.53 GB Insufficient
Q5_K_S 5.7 Very good 364.8 GB 11.31 GB 376.91 GB Insufficient
Q5_K_M 6.21 Very good 397.22 GB 11.31 GB 409.33 GB Insufficient
Q5_K_XL 6.22 Very good 397.88 GB 11.31 GB 409.99 GB Insufficient
Q6_K 6.71 Excellent 429.09 GB 11.31 GB 441.2 GB Insufficient
Q6_K_XL 7.62 Excellent 487.22 GB 11.31 GB 499.33 GB Insufficient
Q8_0 8.51 Excellent 544.13 GB 11.31 GB 556.24 GB Insufficient
Q8_K_XL 8.66 Excellent 553.68 GB 11.31 GB 565.79 GB Insufficient
BF16 16.01 Excellent 1023.56 GB 11.31 GB 1035.67 GB Insufficient

KV cache estimated (architecture unavailable). Speed is a rough estimate bounded by memory bandwidth.

Frequently asked questions

Can I run unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF on an 8 GB GPU?

No. unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF does not fit on an 8 GB GPU, even with the smallest quantization and system RAM offloading.

Can I run unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF on a 16 GB GPU?

No. unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF does not fit on a 16 GB GPU, even with the smallest quantization and system RAM offloading.

Can I run unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF on a 24 GB GPU?

No. unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF does not fit on a 24 GB GPU, even with the smallest quantization and system RAM offloading.

What is the best quantization for unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF?

If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.