What AI models can a GeForce RTX 3080 Ti run?

1931 of 2118 indexed open-weight models fit a GeForce RTX 3080 Ti at 16,384 context with q8_0 KV cache, the largest being c4ai-command-r-08-2024 at Q3_K_L. That covers text, vision-language, image, video and speech models.

How much usable memory does a GeForce RTX 3080 Ti actually have?

Its nameplate is 20 GB, but about 18.60 GiB is available to a model once driver and compositor overhead is accounted for.

Is a GeForce RTX 3080 Ti fast for local AI?

Its memory bandwidth is 760 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 · consumer

GeForce RTX 3080 Ti

GeForce RTX 3080 Ti has 20 GB of VRAM at 760 GB/s — about 18.60 GiB usable after driver and compositor overhead. 1931 of 2118 indexed models fit at 16K context with q8_0 KV.

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

Memory

20 GB

GDDR6X

Bandwidth

760 GB/s

320-bit bus

Tensor FP16

136 TF

dense

TDP

350 W

$1199 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

text 1656vision language 171video 16audio asr 39image 2audio tts 21embedding 26

What fits at 16K context

largest quantization that fits, per model · 1931 of 2118 indexed

Model	Best quant	Params	Weights●	KV●	Total in memory◐	Headroom◐	tok/s◐
c4ai-command-r-08-2024	Q3_K_L	32.3B	16.36 GiB	1.33 GiB	18.60 GiB	0.00 GiB	31±12.9%
t5-v1_1-xxl	F32	4.8B	17.74 GiB	0.00 GiB	18.59 GiB	0.01 GiB	31±12.9%
Trinity-MiniMoE	Q5_K_L	26.1B	17.59 GiB	0.20 GiB	18.58 GiB	0.02 GiB	126±37%
Hy-MT2-30B-A3BMoE	Q4_K_M	30.1B	16.98 GiB	0.80 GiB	18.58 GiB	0.02 GiB	104±37%
Delphi-25B-SimpleRL-Math	I1-Q2_K	25.0B	8.79 GiB	8.89 GiB	18.56 GiB	0.04 GiB	31±12.9%
Rocinante-XL-16B-v1	Q8_0	16.1B	15.91 GiB	1.79 GiB	18.55 GiB	0.05 GiB	31±12.9%
granite-4.1-8b	BF16	8.8B	16.38 GiB	1.33 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Gembrain-X-Core-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Gembrain-X-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-31B-Isometry-Fabled-Persona	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Novelist-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
G4-MeroMero-31B-uncensored-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
G4-Alice-v1.2-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Gemsicle-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Melinoe-Gemma4-31B-VL-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
G4-MeroMero-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Melinoe-Gemma4-31B-VL	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-31B-Storymaxxed3	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Harmonia-31B-uncensored-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Huihui-gemma-4-31B-it-qat-q4_0-unquantized-abliterated	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-Queen-it-qat-q4_0-unquantized	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-AssGuard-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
copywriter-gemma4-31b	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Garnet-V2-31B-it-ultra-uncensored-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-abliterated-v3	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-noloop	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Lilith-31B-v1.0	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Webs-Sejong-31B-v7	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
JGOS-31B-Think	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-Mergemaxxed	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Queen-31B-it-uncensored-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Sphinsikus-Chronist-31B	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma4-31B-Finetuned-V2	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-31B-storymaxxed2	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-Grand-Horror-X-INTENSE-HERETIC-UNCENSORED-Thinking	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-uncensored-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Huihui-gemma-4-31B-it-abliterated-v2	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-Queen-31B-it	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-abliterated	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31b-it-heretic-ara	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-it-heretic	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Gemma-4-31B-Fable-Coder	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B-anthology	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Omni-31B-Turkish-Reasoning-Model	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31b-kairos	IQ4_XS	31.3B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
gemma-4-31B	IQ4_XS	32.7B	15.70 GiB	1.95 GiB	18.54 GiB	0.06 GiB	31±12.9%
Huihui-Qwen3.5-35B-A3B-abliteratedMoE	IQ4_XS	36.0B	17.57 GiB	0.17 GiB	18.54 GiB	0.06 GiB	162±37%
Qwen3.5-35B-A3B-BaseMoE	IQ4_XS	36.0B	17.57 GiB	0.17 GiB	18.54 GiB	0.06 GiB	162±37%
GLM-4.7-Flash-hereticMoE	Q4_K_M	29.9B	17.27 GiB	0.44 GiB	18.52 GiB	0.08 GiB	117±37%
GRM-2.6-Plus-0628	Q4_K_M	27.8B	17.12 GiB	0.53 GiB	18.51 GiB	0.09 GiB	31±12.9%
EXAONE-4.5-33B	I1-IQ4_XS	34.4B	16.63 GiB	0.98 GiB	18.51 GiB	0.09 GiB	31±12.9%
Qwen3-53B-A3B-2507-THINKING-TOTAL-RECALL-v2-MASTER-CODERMoE	I1-IQ2_M	53.0B	16.32 GiB	1.39 GiB	18.51 GiB	0.09 GiB	84±37%
Goetia-26B-A4B-v1.4MoE	I1-Q5_K_S	26.0B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
G4-Moonlight-Dusk-26B-A4B-hereticMoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
Pantheon-Reasoning-26B-A4B-1.1-hereticMoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
G4-Moonlight-Dusk-26B-A4BMoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
Chimera-X-26B-A4BMoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
Pantheon-Reasoning-26B-A4B-1.1MoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%
Gemma-4-26B-A4B-StyleTune-V2MoE	I1-Q5_K_S	26.5B	17.22 GiB	0.49 GiB	18.50 GiB	0.10 GiB	31±12.9%

●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 GeForce RTX 3080 Ti run?: 1931 of 2118 indexed open-weight models fit a GeForce RTX 3080 Ti at 16,384 context with q8_0 KV cache, the largest being c4ai-command-r-08-2024 at Q3_K_L. That covers text, vision-language, image, video and speech models.
How much usable memory does a GeForce RTX 3080 Ti actually have?: Its nameplate is 20 GB, but about 18.60 GiB is available to a model once driver and compositor overhead is accounted for.
Is a GeForce RTX 3080 Ti fast for local AI?: Its memory bandwidth is 760 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.