What AI models can a Titan V run?

1792 of 2118 indexed open-weight models fit a Titan V at 4,096 context with q8_0 KV cache, the largest being Noromaid-20b-v0.1.1 at Q2_K. That covers text, vision-language, image, video and speech models.

How much usable memory does a Titan V actually have?

Its nameplate is 12 GB, but about 11.16 GiB is available to a model once driver and compositor overhead is accounted for.

Is a Titan V fast for local AI?

Its memory bandwidth is 651 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

Titan V

Titan V has 12 GB of VRAM at 651 GB/s — about 11.16 GiB usable after driver and compositor overhead. 1792 of 2118 indexed models fit at 4K context with q8_0 KV.

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

Memory

12 GB

HBM2

Bandwidth

651 GB/s

3072-bit bus

Tensor FP16

119 TF

dense

TDP

250 W

$2999 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 1541video 14vision language 149audio asr 39audio tts 21image 2embedding 26

What fits at 4K context

largest quantization that fits, per model · 1792 of 2118 indexed

Model	Best quant	Params	Weights●	KV●	Total in memory◐	Headroom◐	tok/s◐
Noromaid-20b-v0.1.1	Q2_K	20.0B	7.74 GiB	2.57 GiB	11.16 GiB	0.00 GiB	45±12.9%
Nethena-20B	Q2_K	20.0B	7.74 GiB	2.57 GiB	11.16 GiB	0.00 GiB	45±12.9%
Wan2.1-FLF2V-14B-720P	Q4_1	16.4B	10.32 GiB	0.00 GiB	11.16 GiB	0.00 GiB	45±12.9%
GLM-4.7-FlashMoE	UD-IQ2_M	31.2B	10.24 GiB	0.11 GiB	11.16 GiB	0.00 GiB	179±37%
Wan2.1-I2V-14B-480P	Q4_1	16.4B	10.32 GiB	0.00 GiB	11.15 GiB	0.01 GiB	45±12.9%
Wan2.1-I2V-14B-720P	Q4_1	16.4B	10.32 GiB	0.00 GiB	11.15 GiB	0.01 GiB	45±12.9%
medgemma-27b-it	I1-Q2_K	28.8B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
gemma-3-27b-it-abliterated-refined-vision	I1-Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
gemma-3-27b-it-abliterated	Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
Nidum-Gemma-3-27B-it-Uncensored	I1-Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
gemma-3-27b-it	Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
AtomicGPT-gemma3-27b	I1-Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
Unbound-v1.12.0-27B	I1-Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
Mira-v1.12-Ties-27B	I1-Q2_K	27.4B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
Medgamma27B	I1-Q2_K	27.0B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
medgemma-27b-text-it	Q2_K	27.0B	9.78 GiB	0.49 GiB	11.15 GiB	0.01 GiB	45±12.9%
MiroThinker-v1.0-30BMoE	Q2_K	30.5B	10.16 GiB	0.20 GiB	11.15 GiB	0.01 GiB	170±37%
Pantheon-Proto-RP-1.8-30B-A3BMoE	Q2_K	30.5B	10.16 GiB	0.20 GiB	11.15 GiB	0.01 GiB	170±37%
Phi-4-reasoning	Q5_K_M	14.7B	9.88 GiB	0.42 GiB	11.15 GiB	0.01 GiB	45±12.9%
Phi-4-reasoning-plus	Q5_K_M	14.7B	9.88 GiB	0.42 GiB	11.15 GiB	0.01 GiB	45±12.9%
phi-4	Q5_K_M	14.7B	9.88 GiB	0.42 GiB	11.15 GiB	0.01 GiB	45±12.9%
Tongyi-DeepResearch-30B-A3BMoE	Q2_K	30.5B	10.16 GiB	0.20 GiB	11.15 GiB	0.01 GiB	170±37%
MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking	I1-Q3_K_S	23.4B	9.63 GiB	0.67 GiB	11.15 GiB	0.01 GiB	45±12.9%
Goetia-26B-A4B-v1.4MoE	I1-Q2_K_S	26.0B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
G4-Moonlight-Dusk-26B-A4B-hereticMoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Pantheon-Reasoning-26B-A4B-1.1-hereticMoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
G4-Moonlight-Dusk-26B-A4BMoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Chimera-X-26B-A4BMoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Pantheon-Reasoning-26B-A4B-1.1MoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Gemma-4-26B-A4B-StyleTune-V2MoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Gemma-4-26B-A4B-StyleTuneMoE	I1-Q2_K_S	26.5B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
gemma-4-26b-a4b-heretic-styletune-v2-headMoE	I1-Q2_K_S	25.8B	10.12 GiB	0.24 GiB	11.15 GiB	0.01 GiB	45±12.9%
Gemma-4-31B-Isometry-RP	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Gemma-4-Dark-Gemistry-31B	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Prosopon-31B	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Gemma-4-Novelist-Eclipse-31B	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Giftige-Blume-31B-v1-StyleSwap	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
G4-MeroMero-31B-StyleSwap	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Gemma-4-31B-StyleTune-heretic-ara	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Pantheon-Reasoning-31B-1.1	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Gemma-4-31B-StyleTune	I1-IQ2_XS	32.7B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Barcenas-StyleTune-31B-Fable	I1-IQ2_XS	32.1B	9.31 GiB	0.95 GiB	11.14 GiB	0.02 GiB	45±12.9%
Rocinante-XL-16B-v1	Q4_K_L	16.1B	9.84 GiB	0.45 GiB	11.13 GiB	0.03 GiB	45±12.9%
Ornith-1.0-35BMoE	UD-IQ1_M	34.7B	10.29 GiB	0.04 GiB	11.13 GiB	0.03 GiB	235±37%
ThinkingCap-Qwen3.6-27B	IQ2_M	27.4B	10.13 GiB	0.13 GiB	11.12 GiB	0.04 GiB	45±12.9%
Tess-4-27B	IQ2_M	27.8B	10.13 GiB	0.13 GiB	11.12 GiB	0.04 GiB	45±12.9%
Trinity-MiniMoE	IQ3_XS	26.1B	10.23 GiB	0.10 GiB	11.12 GiB	0.04 GiB	180±37%
Qwen3.5-40B-RoughHouse-Claude-4.6-Opus-Polar-Deckard-Uncensored-Heretic-Thinking	I1-IQ1_M	39.5B	10.05 GiB	0.20 GiB	11.12 GiB	0.04 GiB	45±12.9%
Skyfall-31B-v4.2	IQ2_XS	31.4B	9.75 GiB	0.45 GiB	11.12 GiB	0.04 GiB	45±12.9%
Pantheon-Reasoning-27B	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwen3.6-27B-uncensored-heretic-v2-Native-MTP-Preserved	I1-Q2_K	27.4B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwen3.6-27B-Fable-5-Experimental	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwable-5-27B-Coder	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
EVE-27b-XENO-HAT-DeepSeek-V4-Flash	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
EVE-27B-XENO-HAT	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Godoter-27B	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Reasoning-Medical-27B	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwopus3.6-27B-v2-abliterated	I1-Q2_K	27.4B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±12.9%
Qwen3.6-27B-Uncensored-HauhauCS-Aggressive-BF16	I1-Q2_K	27.8B	10.12 GiB	0.13 GiB	11.11 GiB	0.05 GiB	45±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 Titan V run?: 1792 of 2118 indexed open-weight models fit a Titan V at 4,096 context with q8_0 KV cache, the largest being Noromaid-20b-v0.1.1 at Q2_K. That covers text, vision-language, image, video and speech models.
How much usable memory does a Titan V actually have?: Its nameplate is 12 GB, but about 11.16 GiB is available to a model once driver and compositor overhead is accounted for.
Is a Titan V fast for local AI?: Its memory bandwidth is 651 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.