What AI models can a RTX 5000 Ada Generation run?

2019 of 2118 indexed open-weight models fit a RTX 5000 Ada Generation at 16,384 context with q4_0 KV cache, the largest being Bernini-R at Q8_0. That covers text, vision-language, image, video and speech models.

How much usable memory does a RTX 5000 Ada Generation actually have?

Its nameplate is 32 GB, but about 29.76 GiB is available to a model once driver and compositor overhead is accounted for.

Is a RTX 5000 Ada Generation fast for local AI?

Its memory bandwidth is 576 GB/s, and that figure — not teraflops — is what governs token generation speed. Capacity decides what you can run; bandwidth decides how fast it runs.

NVIDIA · workstation

RTX 5000 Ada Generation

RTX 5000 Ada Generation has 32 GB of VRAM at 576 GB/s — about 29.76 GiB usable after driver and compositor overhead. 2019 of 2118 indexed models fit at 16K context with q4_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed

Memory

32 GB

GDDR6

Bandwidth

576 GB/s

256-bit bus

Tensor FP16

261 TF

dense

TDP

250 W

$4000 MSRP

Context4K 8K 16K 32K 64K 128K

KV cachef16 q8_0 q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context

video 16vision language 181text 1734image 2embedding 26audio tts 21audio asr 39

What fits at 16K context

largest quantization that fits, per model · 2019 of 2118 indexed

Model	Best quant	Params	Weights●	KV●	Total in memory◐	Headroom◐	tok/s◐
Bernini-R	Q8_0	14.3B	28.71 GiB	0.00 GiB	29.76 GiB	0.00 GiB	12±22%
Salience-1.5-ProMoE	Q6_K_L	36.0B	28.66 GiB	0.09 GiB	29.75 GiB	0.01 GiB	69±37%
Qwable-v1MoE	Q6_K_L	36.0B	28.66 GiB	0.09 GiB	29.75 GiB	0.01 GiB	69±37%
T-SearchMoE	Q6_K_L	36.0B	28.66 GiB	0.09 GiB	29.75 GiB	0.01 GiB	69±37%
Melody1437-27B	Q3_K_M	27.8B	28.40 GiB	0.28 GiB	29.75 GiB	0.01 GiB	12±22%
Qwen3-53B-A3B-2507-THINKING-TOTAL-RECALL-v2-MASTER-CODERMoE	I1-Q4_0	53.0B	28.02 GiB	0.74 GiB	29.75 GiB	0.01 GiB	41±37%
Gemma-3-27B-MeditronFO	Q8_0	28.8B	28.13 GiB	0.52 GiB	29.73 GiB	0.03 GiB	12±22%
Qwen2.5-7B-Instruct-1M	F32	7.6B	28.38 GiB	0.25 GiB	29.68 GiB	0.08 GiB	12±22%
DeepSeek-R1-Distill-Qwen-7B	F32	7.6B	28.38 GiB	0.25 GiB	29.68 GiB	0.08 GiB	12±22%
UI-TARS-7B-DPO	F32	8.3B	28.38 GiB	0.25 GiB	29.68 GiB	0.08 GiB	12±22%
Qwen2-7B-Instruct	F32	7.6B	28.38 GiB	0.25 GiB	29.68 GiB	0.08 GiB	12±22%
Hercules-5.0-Qwen2-7B	F32	7.6B	28.38 GiB	0.25 GiB	29.68 GiB	0.08 GiB	12±22%
Kepler-8B-Instruct-v2	F16	7.6B	28.37 GiB	0.25 GiB	29.67 GiB	0.09 GiB	12±22%
MiniCPM-o-2_6	F32	8.7B	28.37 GiB	0.25 GiB	29.67 GiB	0.09 GiB	12±22%
Mixtral-8x22B-Instruct-v0.1MoE	IQ1_S	141B	27.61 GiB	0.98 GiB	29.66 GiB	0.10 GiB	21±37%
Mixtral-8x22B-v0.1MoE	IQ1_S	141B	27.61 GiB	0.98 GiB	29.65 GiB	0.11 GiB	21±37%
CalmeRys-78B-Orpo-v0.1	I1-IQ2_XS	78.0B	27.00 GiB	1.51 GiB	29.64 GiB	0.12 GiB	12±22%
calme-2.3-rys-78b	IQ2_XS	78.0B	27.00 GiB	1.51 GiB	29.64 GiB	0.12 GiB	12±22%
Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODERMoE	I1-Q5_K_M	42.4B	28.05 GiB	0.59 GiB	29.64 GiB	0.12 GiB	47±37%
Huihui-GLM-4.7-Flash-abliterated-57BMoE	IQ4_XS	57.3B	27.93 GiB	0.59 GiB	29.56 GiB	0.20 GiB	47±37%
Assistant_Pepe_70B	IQ3_XXS	70.6B	27.03 GiB	1.41 GiB	29.56 GiB	0.20 GiB	12±22%
Qwen3.5-88BMoE	I1-Q2_K_S	87.7B	28.30 GiB	0.11 GiB	29.43 GiB	0.33 GiB	61±37%
DeepCoder-14B-Preview	BF16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
SuperNova-Medius	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-14B-Instruct-1M	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
OpenCodeReasoning-Nemotron-14B	BF16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-Coder-14B-Instruct-abliterated	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
0x-lite	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-Coder-14B-Instruct	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-14B-Instruct-1M-abliterated	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
AceReason-Nemotron-14B	BF16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-14B-Instruct	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Qwen2.5-Coder-14B	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
DeepSeek-R1-Distill-Qwen-14B-abliterated-v2	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
DeepSeek-R1-Distill-Qwen-14B	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Sugoi-14B-Ultra-HF	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
UwU-14B-Math-v0.2	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
oxy-1-small	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
EVA-Qwen2.5-14B-v0.2	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
EVA-Qwen2.5-14B-v0.0	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
EVA-Qwen2.5-14B-v0.1	F16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Strand-Rust-Coder-14B-v1	BF16	14.8B	27.52 GiB	0.84 GiB	29.41 GiB	0.35 GiB	12±22%
Lamarck-14B-v0.7	F16	14.8B	27.51 GiB	0.84 GiB	29.40 GiB	0.36 GiB	12±22%
command-r-35b-writer-v2	I1-Q5_K_S	35.0B	22.67 GiB	5.63 GiB	29.40 GiB	0.36 GiB	12±22%
Kimi-Linear-48B-A3B-InstructMoE	Q4_K_L	49.1B	28.26 GiB	0.13 GiB	29.40 GiB	0.36 GiB	12±22%
WizardLM-Uncensored-SuperCOT-StoryTelling-30b	Q5_K_M	32.5B	21.46 GiB	6.86 GiB	29.39 GiB	0.37 GiB	12±22%
Wizard-Vicuna-30B-Uncensored	I1-Q5_K_M	32.5B	21.46 GiB	6.86 GiB	29.39 GiB	0.37 GiB	12±22%
archangel_sft-kto_llama30b	I1-Q5_K_M	32.5B	21.46 GiB	6.86 GiB	29.39 GiB	0.37 GiB	12±22%
Phi-3.5-MoE-instructMoEKV unresolved	Q5_K_L	41.9B	27.75 GiB	0.56 GiB	29.32 GiB	0.44 GiB	34±37%
Darwin-35B-A3B-OpusMoE	Q6_K_L	36.0B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
Aurora-Code-1MoE	Q6_K_L	34.7B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
grug-35b-v2MoE	Q6_K_L	35.1B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
grug-35bMoE	Q6_K_L	35.1B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
WorldSim-Opus-3.6-35B-A3BMoE	Q6_K_L	35.1B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
Qwen3.6-35B-A3B-AnkoMoE	Q6_K_L	35.1B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
KAT-Coder-V2.5-DevMoE	Q6_K_L	34.7B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
Ornith-1.0-35BMoE	Q6_K_L	34.7B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
Nex-N2-miniMoE	Q6_K_L	35.1B	28.22 GiB	0.09 GiB	29.31 GiB	0.45 GiB	70±37%
deepseek-llm-67b-chat	Q2_K	67.4B	26.54 GiB	1.67 GiB	29.31 GiB	0.45 GiB	12±22%
deepseek-llm-67b-base	Q2_K	67.4B	26.54 GiB	1.67 GiB	29.31 GiB	0.45 GiB	12±22%

●From the file◐Predictedwhat these mean

Speed is modeled, not measured: decode is memory-bandwidth bound, so tokens per second is bytes read per token against achievable bandwidth. Mixture-of-experts models carry a wider band because only the routed experts are read each step, and few have been measured publicly.

Questions people ask

What AI models can a RTX 5000 Ada Generation run?: 2019 of 2118 indexed open-weight models fit a RTX 5000 Ada Generation at 16,384 context with q4_0 KV cache, the largest being Bernini-R at Q8_0. That covers text, vision-language, image, video and speech models.
How much usable memory does a RTX 5000 Ada Generation actually have?: Its nameplate is 32 GB, but about 29.76 GiB is available to a model once driver and compositor overhead is accounted for.
Is a RTX 5000 Ada Generation fast for local AI?: Its memory bandwidth is 576 GB/s, and that figure — not teraflops — is what governs token generation speed. Capacity decides what you can run; bandwidth decides how fast it runs.