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. 977 of 2118 indexed models fit at 64K context with f16 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
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 784video 14embedding 22audio asr 38audio tts 20vision language 98image 1

What fits at 64K context

largest quantization that fits, per model · 977 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
OLMoE-1B-7B-0924-InstructMoEI1-Q2_K6.9B2.39 GiB8.00 GiB11.16 GiB0.00 GiB32±37%
Wan2.1-FLF2V-14B-720PQ4_116.4B10.32 GiB0.00 GiB11.16 GiB0.00 GiB45±12.9%
Wan2.1-I2V-14B-480PQ4_116.4B10.32 GiB0.00 GiB11.15 GiB0.01 GiB45±12.9%
Wan2.1-I2V-14B-720PQ4_116.4B10.32 GiB0.00 GiB11.15 GiB0.01 GiB45±12.9%
Nemotron-3-Embed-8B-BF16IQ1_S8.0B1.81 GiB8.50 GiB11.15 GiB0.01 GiB45±12.9%
granite-speech-4.1-2b-narF162.3B5.36 GiB5.00 GiB11.15 GiB0.01 GiB45±12.9%
Ace-Step1.5Q4_K160M7.32 GiB3.04 GiB11.15 GiB0.01 GiB45±12.9%
Muse-Glimmer-30BIQ2_S29.8B9.35 GiB0.91 GiB11.14 GiB0.02 GiB45±12.9%
Phi-4-mini-instruct-abliteratedQ4_K_M3.8B2.32 GiB8.00 GiB11.13 GiB0.03 GiB45±12.9%
Phi-4-mini-reasoningQ4_K_M3.8B2.32 GiB8.00 GiB11.13 GiB0.03 GiB45±12.9%
Phi-4-mini-instructQ4_K_M3.8B2.32 GiB8.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-VL-4B-Instruct-Unredacted-MAXI1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-VL-4B-Thinking-Unredacted-MAXI1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Zubr1.0-VL-4BI1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Huihui-Qwen3-VL-4B-Instruct-abliteratedI1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-VL-4B-Instruct-UncensoredI1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
OpenCaption-4B-VL-SFT-v1.0I1-IQ2_S4.4B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Parable-Qwen3-4B-Claude-Fable-5I1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-4b-Z-Image-Turbo-AbliteratedV1I1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Neuron-4B-InstructI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
ChineseErrorCorrector4-4BI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
FastContext-1.0-4B-SFT-abliteratedI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-4B-Instruct_NSFW-V2.1I1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
FastContext-1.0-4B-SFTI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
fable-traces-abliteratedI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Nexa-AI-4B-InstructI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Lumen-4B-InstructI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Qwen3-HereticLM-4BI1-IQ2_S4.0B1.32 GiB9.00 GiB11.13 GiB0.03 GiB45±12.9%
Marco-Nano-InstructMoEI1-IQ3_XXS8.0B3.35 GiB7.00 GiB11.12 GiB0.04 GiB36±37%
legitus-instruct-v1I1-IQ2_XXS8.1B2.25 GiB8.00 GiB11.12 GiB0.04 GiB45±12.9%
Apertus-8B-Instruct-2509I1-IQ2_XXS8.1B2.25 GiB8.00 GiB11.12 GiB0.04 GiB45±12.9%
Nemotron-Mini-4B-InstructIQ4_XS4.2B2.29 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
Hubble-4B-v1Q3_K_L4.5B2.30 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
Aura-4BI1-Q3_K_L4.5B2.30 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
magnum-v2-4bI1-Q3_K_L4.5B2.30 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
Impish_LLAMA_4BQ3_K_L4.5B2.30 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
Llama-3.1-Minitron-4B-Width-BaseQ3_K_L4.5B2.30 GiB8.00 GiB11.11 GiB0.05 GiB45±12.9%
Wan2.2-Distill-ModelsQ5_114.3B10.27 GiB0.00 GiB11.10 GiB0.06 GiB45±12.9%
Marco-Mini-InstructMoEI1-IQ1_S17.3B3.33 GiB7.00 GiB11.10 GiB0.06 GiB36±37%
Hunyuan-7B-InstructIQ2_XS7.5B2.26 GiB8.00 GiB11.10 GiB0.06 GiB45±12.9%
Bernini-RQ5_114.3B10.26 GiB0.00 GiB11.10 GiB0.06 GiB45±12.9%
SkyReels-V2-DF-14B-540PQ5_114.3B10.27 GiB0.00 GiB11.10 GiB0.06 GiB45±12.9%
CycleGRPO-4BI1-IQ2_XXS4.8B1.28 GiB9.00 GiB11.09 GiB0.07 GiB45±12.9%
Voxtral-Mini-3B-2507Q4_K_M4.7B2.78 GiB7.50 GiB11.09 GiB0.07 GiB45±12.9%
InternVL3_5-14BQ5_K_L15.1B10.24 GiB0.00 GiB11.08 GiB0.08 GiB45±12.9%
orpheus-3b-0.1-ftQ8_03.8B3.27 GiB7.00 GiB11.08 GiB0.08 GiB45±12.9%
Ministral-8B-Instruct-2410Q4_K_M8.0B4.57 GiB5.68 GiB11.08 GiB0.08 GiB45±12.9%
Foundation-Sec-8B-InstructI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Foundation-Sec-8B-Instruct-hereticI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Meta-Llama-3-8BIQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
llama3.1-heretic-unsensoredI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Anubis-Mini-8B-v1I1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
dolphin-2.9-llama3-8bIQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Meta-Llama-3-8BIQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8B-hereticI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8BI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Gluon-8BI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-ReasoningI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Llama3.3-8B-Instruct-Thinking-Claude-4.5-Opus-High-ReasoningI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±12.9%
Hypnos-i1-8BI1-IQ2_XXS8.0B2.23 GiB8.00 GiB11.07 GiB0.09 GiB45±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?
977 of 2118 indexed open-weight models fit a Titan V at 65,536 context with f16 KV cache, the largest being OLMoE-1B-7B-0924-Instruct at I1-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.
Titan V — what AI models can it run locally? — ossmodeldb