GPUs & Macs › NVIDIA RTX 6000 Ada

NVIDIA RTX 6000 Ada: local LLM speed

Llama 2 7B (Q4_0) generates 179 tokens/s on the NVIDIA RTX 6000 Ada (CUDA), and processes prompts at 10,577 tokens/s.

Measured results

ModelGenerate
tok/s		Prompt
tok/s		QuantBackendTestSource
Llama 2 7B17910,577Q4_0CUDA FA ontg128llama.cpp #15013 details
Llama 2 7B1769,229Q4_0CUDA FA offtg128llama.cpp #15013 details
Llama 3.2 1B Instruct27123,450Q4_K_Mllamafile LocalScore avg · n=2LocalScore LocalScore details
Llama 3.1 8B Instruct86.16,148Q4_K_Mllamafile LocalScore avg · n=2LocalScore LocalScore details
Qwen2.5 14B Instruct49.13,244Q4_K_Mllamafile LocalScore avg · n=2LocalScore LocalScore details

What fits on the NVIDIA RTX 6000 Ada

Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 179 tok/s (CUDA, source). Fit assumes the whole model in GPU memory with ~8K context.

ModelBest quant that fitsFileExpected tok/s
Qwen3.8-27BQ8_029.0 GB23.6 estimate
gemma-4-12B-itQ8_012.7 GB54.1 estimate
Qwen3.5-9BQ8_09.5 GB72.0 estimate
Qwen3.5-4BQ8_04.5 GB153 estimate
Qwen3.6-35B-A3BQ8_036.9 GB108 estimate
gemma-4-26B-A4B-itQ8_026.9 GB83.0 estimate
MiMo-V2.6-Distill-Qwen-9BQ8_09.5 GB71.9 estimate
gemma-4-E4B-itQ8_08.2 GB83.7 estimate
Muse-Glimmer-30BQ8_029.6 GB23.2 estimate
Qwen3-4BQ8_04.3 GB160 estimate
gemma-4-31B-itQ8_032.6 GB21.0 estimate
gemma-4-26B-A4B-it-qat-q4_0-unquantizedUD-Q4_K_XL14.2 GB157 estimate
gemma-4-12B-it-qat-q4_0-unquantizedUD-Q4_K_XL6.7 GB102 estimate
gemma-3-1b-itQ8_01.1 GB572 estimate

Open in the picker

Similar hardware

NVIDIA RTX A6000 · NVIDIA Quadro RTX 8000 · NVIDIA RTX PRO 4500 Blackwell · NVIDIA RTX PRO 4000 Blackwell · NVIDIA RTX A5000 · NVIDIA Quadro RTX 6000 · NVIDIA RTX A4500 · NVIDIA RTX 4000 Ada