NVIDIA · consumer

GeForce RTX 3050

GeForce RTX 3050 has 6 GB of VRAM at 168 GB/s — about 5.58 GiB usable after driver and compositor overhead. 1105 of 2118 indexed models fit at 32K context with q4_0 KV.

Spec sheet· bandwidth, theoreticalFrom the file· fit from summed bytesPredicted· speed
Memory
6 GB
GDDR6
Bandwidth
168 GB/s
96-bit bus
Tensor FP16
27 TF
dense
TDP
70 W
$179 MSRP
Context4K8K16K32K64K128K
KV cachef16q8_0q4_0quantizing the KV cache is a ~2× lever on the dominant term at long context
text 931audio asr 38audio tts 19embedding 26vision language 88video 3

What fits at 32K context

largest quantization that fits, per model · 1105 of 2118 indexed
ModelBest quantParamsWeights●KV●Total in memory◐Headroom◐tok/s◐
Aya-Medikal-V2I1-Q3_K_S8.0B3.60 GiB1.13 GiB5.57 GiB0.01 GiB26±12.9%
canary-qwen-2.5bBF162.6B4.73 GiB0.00 GiB5.57 GiB0.01 GiB26±12.9%
EXAONE-Deep-7.8BQ4_K_L7.8B4.73 GiB0.00 GiB5.57 GiB0.01 GiB26±12.9%
EXAONE-3.5-7.8B-InstructQ4_K_L7.8B4.73 GiB0.00 GiB5.57 GiB0.01 GiB26±12.9%
LFM2.5-8B-A1BMoEUD-Q4_K_S8.5B4.67 GiB0.11 GiB5.57 GiB0.01 GiB71±37%
Hubble-4B-v1Q6_K_L4.5B3.63 GiB1.13 GiB5.57 GiB0.01 GiB25±12.9%
Aura-4BQ6_K_L4.5B3.63 GiB1.13 GiB5.57 GiB0.01 GiB25±12.9%
magnum-v2-4bQ6_K_L4.5B3.63 GiB1.13 GiB5.57 GiB0.01 GiB25±12.9%
Impish_LLAMA_4BQ6_K_L4.5B3.63 GiB1.13 GiB5.57 GiB0.01 GiB25±12.9%
Llama-3.1-Minitron-4B-Width-BaseQ6_K_L4.5B3.63 GiB1.13 GiB5.57 GiB0.01 GiB25±12.9%
OLMoE-1B-7B-0924-InstructMoEI1-Q4_06.9B3.67 GiB1.13 GiB5.57 GiB0.01 GiB38±37%
Qwen3-TTS-12Hz-0.6B-BaseQ4_K_M915M4.72 GiB0.00 GiB5.57 GiB0.01 GiB26±12.9%
VoxCPM2F162.3B4.72 GiB0.00 GiB5.57 GiB0.01 GiB26±12.9%
Qwen3-1.7BBF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
DualMinded-Qwen3-1.7BF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
DualMindF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
Qwen3-1.7B-Coder-Distilled-SFTF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
Qwen3-1.7B-Distilled-30B-A3B-SFTF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
DistilQwen3-1.7B-uncensoredF162.0B3.79 GiB0.98 GiB5.57 GiB0.01 GiB25±12.9%
nomic-embed-codeQ4_17.1B4.22 GiB0.49 GiB5.57 GiB0.01 GiB26±12.9%
gemma-4-E2B-itQ8_05.1B4.70 GiB0.07 GiB5.56 GiB0.02 GiB25±12.9%
gemma-4-E2B-itQ8_05.1B4.70 GiB0.07 GiB5.56 GiB0.02 GiB25±12.9%
t5-v1_1-xxlQ2_K4.8B4.72 GiB0.00 GiB5.56 GiB0.02 GiB26±12.9%
Tini-Cybersec-8B-A1BMoEQ4_K_S8.5B4.66 GiB0.11 GiB5.56 GiB0.02 GiB71±37%
L3-Dark-Planet-8BQ3_K_S8.0B3.60 GiB1.13 GiB5.56 GiB0.02 GiB26±12.9%
salamandra-7b-instruct-2606I1-IQ3_M7.8B3.60 GiB1.13 GiB5.56 GiB0.02 GiB26±12.9%
Phi-3.5-mini-instructIQ3_XXS3.8B1.37 GiB3.38 GiB5.55 GiB0.03 GiB26±12.9%
gemma-3-12b-it-vl-Gemini-3-Pro-Preview-Heretic-Uncensored-ThinkingI1-IQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
gemma-3-12b-it-vl-Deepseek-v3.1-Heretic-Uncensored-ThinkingI1-IQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
gemma-3-12b-it-vl-GLM-4.7-Flash-Heretic-Uncensored-ThinkingI1-IQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
Floppa-12B-Gemma3-UncensoredI1-IQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
gemma-3-12b-it-hereticI1-IQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
gemma-3-12b-it-abliteratedIQ2_M12.2B4.01 GiB0.69 GiB5.55 GiB0.03 GiB26±12.9%
phi-2Q5_K_M2.8B1.93 GiB2.81 GiB5.55 GiB0.03 GiB26±12.9%
Ministral-8B-Instruct-2410Q2_K_L8.0B3.45 GiB1.27 GiB5.55 GiB0.03 GiB26±12.9%
Luna-7B-A4BMoEIQ4_XS6.7B3.48 GiB1.27 GiB5.55 GiB0.03 GiB22±37%
Jan-code-4bQ6_K4.4B3.47 GiB1.27 GiB5.55 GiB0.03 GiB26±12.9%
orpheus-3b-0.1-pretrainedQ8_03.8B3.75 GiB0.98 GiB5.55 GiB0.03 GiB26±12.9%
gemma-4-E2B-it-ultra-uncensored-hereticQ8_05.1B4.68 GiB0.07 GiB5.55 GiB0.03 GiB25±12.9%
legitus-instruct-v1I1-IQ3_M8.1B3.55 GiB1.13 GiB5.55 GiB0.03 GiB26±12.9%
Apertus-8B-Instruct-2509I1-IQ3_M8.1B3.55 GiB1.13 GiB5.55 GiB0.03 GiB26±12.9%
Falcon3-7B-InstructQ3_K_L7.5B3.70 GiB0.98 GiB5.55 GiB0.03 GiB26±12.9%
Nemotron-3-Embed-8B-BF16IQ3_M8.0B3.50 GiB1.20 GiB5.54 GiB0.04 GiB26±12.9%
SOLAR-10.7B-Instruct-v1.0I1-IQ2_XS10.7B3.01 GiB1.69 GiB5.53 GiB0.05 GiB26±12.9%
granite-8b-code-instruct-4kI1-IQ3_M8.1B3.43 GiB1.27 GiB5.53 GiB0.05 GiB26±12.9%
granite-8b-code-base-4kI1-IQ3_M8.1B3.43 GiB1.27 GiB5.53 GiB0.05 GiB26±12.9%
Qwen3.5-9B-CoderI1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Qwopus3.5-9B-v3.5Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Qwythos-9B-Claude-Mythos-5-1M-MTPI1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Huihui-Qwythos-9B-Claude-Mythos-5-1M-abliteratedI1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Qwen3.5-9B-Fable-5-v1I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Qwythos-9B-v2I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
PINQWEN-3.5-9B-1M-BF16I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Openprose-2-FlashI1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Qwen3.5-9B-Nikusui-v1I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Ornstein-3.5-9B-V1.5I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Ornith-1.0-9B-heretic-MTPI1-Q3_K_M9.4B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
Tess-4-9BI1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
dotwebs-1I1-Q3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±12.9%
liftQ3_K_M9.7B4.41 GiB0.28 GiB5.53 GiB0.05 GiB26±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.

Measured on this card

third-party benchmarks, aggregated
Workload◍MedianMiddle 50%Runs
Image generation0.31 it/s0.21–2.479
Benchmarked· n=9

Aggregated from community-submitted runs, so the spread is wide by nature — it covers different models, resolutions, step counts and settings, not one controlled configuration. Read the middle 50% rather than the median alone. These figures are reproduced with attribution from vladmandic-sd-data-benchmark, which publishes no licence — so we display and link rather than redistribute them.

Questions people ask

What AI models can a GeForce RTX 3050 run?
1105 of 2118 indexed open-weight models fit a GeForce RTX 3050 at 32,768 context with q4_0 KV cache, the largest being Aya-Medikal-V2 at I1-Q3_K_S. That covers text, vision-language, image, video and speech models.
How much usable memory does a GeForce RTX 3050 actually have?
Its nameplate is 6 GB, but about 5.58 GiB is available to a model once driver and compositor overhead is accounted for.
Is a GeForce RTX 3050 fast for local AI?
Its memory bandwidth is 168 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 3050 — what AI models can it run locally? — ossmodeldb