NVIDIA · consumer

GeForce RTX 3080

GeForce RTX 3080 has 10 GB of VRAM at 760 GB/s — about 9.30 GiB usable after driver and compositor overhead. 1681 of 2118 indexed models fit at 8K context with q4_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
10 GB
GDDR6X
Bandwidth
760 GB/s
320-bit bus
Tensor FP16
119 TF
dense
TDP
320 W
$699 MSRP
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 1456vision language 125image 2video 12audio tts 21embedding 26audio asr 39

What fits at 8K context

largest quantization that fits, per model · 1681 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
OmniAtlas-Qwen3-30B-A3BI1-IQ2_XS31.7B8.45 GiB0.00 GiB9.30 GiB0.00 GiB63±12.9%
Qwen3-Omni-30B-A3B-CaptionerI1-IQ2_XS31.7B8.45 GiB0.00 GiB9.30 GiB0.00 GiB63±12.9%
UncensoredLM-DeepSeek-R1-Distill-Qwen-14BQ4_K_M14.2B8.05 GiB0.40 GiB9.30 GiB0.00 GiB63±12.9%
Rocinante-XL-16B-v1I1-Q3_K_L16.1B7.97 GiB0.47 GiB9.30 GiB0.00 GiB63±12.9%
zeta-2.1Q8_08.3B8.17 GiB0.28 GiB9.29 GiB0.01 GiB63±12.9%
zeta-2Q8_08.3B8.17 GiB0.28 GiB9.29 GiB0.01 GiB63±12.9%
Grug-12BQ5_K_M12.0B8.17 GiB0.27 GiB9.29 GiB0.01 GiB63±12.9%
gemma-4-12B-it-Esper4Q5_K_M12.0B8.17 GiB0.27 GiB9.29 GiB0.01 GiB63±12.9%
gemma-4-12B-itQ5_K_M12.0B8.17 GiB0.27 GiB9.29 GiB0.01 GiB63±12.9%
GLM-4.7-Flash-REAP-23B-A3B-absolute-heresyMoEI1-IQ3_XXS23.0B8.36 GiB0.12 GiB9.29 GiB0.01 GiB211±37%
gemma-4-A4B-98e-v6-coder-itMoEIQ3_XXS20.5B8.33 GiB0.17 GiB9.29 GiB0.01 GiB63±12.9%
MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-ThinkingI1-Q2_K_S23.4B7.73 GiB0.71 GiB9.29 GiB0.01 GiB63±12.9%
Nemotron-Mini-4B-InstructQ4_K_S4.2B8.19 GiB0.28 GiB9.29 GiB0.01 GiB63±12.9%
HomunculusQ5_K_S12.5B8.08 GiB0.35 GiB9.28 GiB0.02 GiB63±12.9%
Falcon3-10B-InstructQ6_K_L10.3B8.06 GiB0.35 GiB9.28 GiB0.02 GiB63±12.9%
Phi-3-medium-128k-instructQ4_K_M14.0B7.98 GiB0.44 GiB9.28 GiB0.02 GiB63±12.9%
Phi-3-medium-4k-instructI1-Q4_K_M14.0B7.98 GiB0.44 GiB9.28 GiB0.02 GiB63±12.9%
Skywork-R1V3-38BIQ2_XXS38.4B8.41 GiB0.00 GiB9.27 GiB0.03 GiB63±12.9%
Fallen-Gemma3-27B-v1IQ2_S27.4B8.18 GiB0.26 GiB9.27 GiB0.03 GiB63±12.9%
granite-3.3-8b-instructQ8_08.2B8.09 GiB0.35 GiB9.27 GiB0.03 GiB63±12.9%
granite-3.2-8b-instructQ8_08.2B8.09 GiB0.35 GiB9.27 GiB0.03 GiB63±12.9%
HunyuanImage-2.1Q3_K_S17.5B8.42 GiB0.00 GiB9.27 GiB0.03 GiB63±12.9%
glm-4-9b-chat-abliteratedQ5_K_L9.4B7.01 GiB1.41 GiB9.26 GiB0.04 GiB63±12.9%
glm-4-9b-chatQ5_K_L9.4B7.01 GiB1.41 GiB9.26 GiB0.04 GiB63±12.9%
next-8bQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
next-ocrQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-GLM-4.7-Flash-Heretic-Uncensored-ThinkingQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Midas-FableAgent-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-Heretic-1.3.0Q8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-ThinkingQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-Instruct-Unredacted-MAXQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Poe-8B-GLM5-Opus4.6-Sonnet4.5-Kimi-Grok-Gemini-3-pro-preview-HERETICQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen-3-VL-8B-Instruct-hereticQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
nsfwcaption-qwen3-vl-8b-v3-safetensorsQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Huihui-Qwen3-VL-8B-Instruct-abliteratedQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-Reranker-8BQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Salience-1-9BQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-InstructQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-Instruct-Uncensored-V2Q8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
GRaPE-2-FlashQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Jan-v2-VL-highQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Jan-v2-VL-medQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
DeepSeek-R1-0528-Qwen3-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Parable-Qwen3-8B-Claude-Fable-5Q8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
ReasonCritic-7BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Finch-8B-KTOQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Finch-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
mythos-9b-unhinged-hereticQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
nsfwvision-qwen3-vl-8b-v3-safetensorsQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
MathSmith-hc-Qwen3-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-VL-8B-Thinking-Unredacted-MAXQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
MiroThinker-v1.0-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
mythos-9b-unhingedQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Maestro1-9BQ8_08.8B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
qwen3-8b-claude-agentic-fable5Q8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Ektome-Qwen3-8B-PristinelyUncensoredQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
mythos-9b-mergedQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Qwen3-8BQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
qwen3-8b-apostateQ8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±12.9%
Josiefied-Qwen3-8B-abliterated-v1Q8_08.2B8.11 GiB0.32 GiB9.26 GiB0.04 GiB63±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 run?
1681 of 2118 indexed open-weight models fit a GeForce RTX 3080 at 8,192 context with q4_0 KV cache, the largest being OmniAtlas-Qwen3-30B-A3B at I1-IQ2_XS. That covers text, vision-language, image, video and speech models.
How much usable memory does a GeForce RTX 3080 actually have?
Its nameplate is 10 GB, but about 9.30 GiB is available to a model once driver and compositor overhead is accounted for.
Is a GeForce RTX 3080 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.
GeForce RTX 3080 — what AI models can it run locally? — ossmodeldb