What can you run on a 12GB graphics card?

With 12GB VRAM you can run 45 of the 101 open-source models we track — up to about 16B parameters at Q4 quantization. The most popular that fits is Qwen3.5-9B, needing 6.8 GB and generating around ~58 tok/s on typical hardware of this size.

Typical machines: RTX 4070 · RTX 5070 · RTX 3060 12GB · RX 7700 XT. After the operating system takes its share, about 12 GB is free for a model.

Having the memory is not the same as having the speed

12GB VRAM decides whether a model loads. How fast it answers is decided by memory bandwidth — how quickly your machine can read the model, which it must do once for every word it writes. Machines with 12GB VRAM span 360–672 GB/s, so the fastest is about 1.9× quicker than the slowest with identical capacity. The table quotes a RTX 4070; pick yours below to see the difference. Why bandwidth and not the processor →

360 GB/sGeForce RTX 3060 (12GB)· NVIDIA published specification: 360 GB/s GDDR6, 192-bit
432 GB/sRadeon RX 7700 XT (12GB)· AMD published specification: 432 GB/s GDDR6, 192-bit
504 GB/sGeForce RTX 4070 (12GB)· NVIDIA published specification: 504.2 GB/s GDDR6X, 192-bit
672 GB/sGeForce RTX 5070 (12GB)· NVIDIA published specification: 672 GB/s GDDR7, 192-bit

Models that run on 12GB VRAM in 2026

ModelSizeMemoryEst. speedDownloadsFeels like
granite-embedding-97m-multilingual-r2Ibm97M655 MBRuns well~3106 tok/s237,787Fast
granite-embedding-311m-multilingual-r2Ibm312M748 MBRuns well~1741 tok/s47,167Fast
LFM2.5-Encoder-230MLiquid Ai230M793 MBRuns well~2362 tok/s13,450Fast
LFM2.5-230MLiquid Ai230M808 MBRuns well~2136 tok/s54,986Fast
ERNIE-4.5-0.3B-PTBaidu AI361M853 MBRuns well~1361 tok/s21,323Fast
nemotron-3.5-asr-streaming-0.6bNVIDIA638M1.1 GBRuns well~661 tok/s1,052,774Fast
Qwen3.5-0.8BAlibaba / Qwen873M1.3 GBRuns well~615 tok/s2,990,867Fast
LFM2.5-1.2B-InstructLiquid Ai1.2B1.4 GBRuns well~448 tok/s555,431Fast
LFM2.5-VL-1.6BLiquid Ai1.6B1.4 GBRuns well~448 tok/s78,655Fast
Nemotron-3-Embed-1B-BF16NVIDIA1.1B1.5 GBRuns well~476 tok/s467,453Fast
cohere-transcribe-arabic-07-2026Cohere2.1B1.9 GBRuns well~263 tok/s51,389Fast
Hy-MT2-1.8BTencent2B1.9 GBRuns well~289 tok/s49,645Fast
Qwen3.5-2BAlibaba / Qwen2.3B2.0 GBRuns well~256 tok/s2,468,876Fast
granite-speech-4.1-2bIbm2.3B2.4 GBRuns well~210 tok/s402,320Fast
LFM2.5-2.6BLiquid Ai2.7B2.5 GBRuns well~196 tok/s77,973Fast
LFM2.5-Audio-1.5BLiquid Ai1.5B2.6 GBRuns well~180 tok/s1,025Fast
Unlimited-OCRBaidu AI3.3B2.7 GBRuns well~168 tok/s2,836,694Fast
LocateAnything-3BNVIDIA3.8B2.8 GBRuns well~155 tok/s463,853Fast
granite-4.1-3bIbm3.4B3.0 GBRuns well~156 tok/s300,243Fast
Shieldstral-1.0-3BMistral AI3.8B3.1 GBRuns well~153 tok/s2,480Fast
Ministral-3-3B-Instruct-2512Mistral AI3.8B3.1 GBRuns well~153 tok/s544,419Fast
Nemotron-Labs-Diffusion-3B-BaseNVIDIA3.8B3.3 GBRuns well~142 tok/s35,110Fast
Cosmos3-EdgeNVIDIA3.9B3.3 GBRuns well~141 tok/s53,138Fast
Nemotron-3.5-Content-SafetyNVIDIA4.3B3.6 GBRuns well~132 tok/s176,924Fast
gemma-4-E2B-itGoogle DeepMind5.1B3.8 GBRuns well~105 tok/s4,002,947Fast
Qwen3.5-4BAlibaba / Qwen4.7B3.8 GBRuns well~120 tok/s6,199,203Fast
Fara1.5-4BMicrosoft4.5B4.0 GBRuns well~114 tok/s4,882Fast
NVIDIA-Nemotron-3-Nano-4B-BF16NVIDIA4B4.1 GBRuns well~115 tok/s712,701Fast
Fara-7BMicrosoft8.3B5.5 GBRuns well~70 tok/s2,192Fast
Hy-MT2-7BTencent8B5.7 GBRuns well~71 tok/s26,629Fast
LFM2.5-8B-A1BLiquid Ai · MoE8.5B5.9 GBRuns well~538 tok/s1B active170,464Fast
Nemotron-3-Embed-8B-BF16NVIDIA8B5.9 GBRuns well~68 tok/s142,381Fast
gemma-4-E4B-itGoogle DeepMind8B6.2 GBRuns well~61 tok/s5,295,850Fast
Nemotron-Labs-Diffusion-8B-BaseNVIDIA8.5B6.2 GBRuns well~64 tok/s413,039Fast
Ministral-3-8B-Instruct-2512Mistral AI8.9B6.3 GBRuns well~63 tok/s119,379Fast
granite-4.1-8bIbm8.8B6.3 GBRuns well~64 tok/s4,213,980Fast
Ornith-1.0-9BDeepreinforce9B6.7 GBRuns well~58 tok/s2,250,422Fast
Qwen3.5-9BAlibaba / Qwen9.7B6.8 GBRuns well~58 tok/s11,767,971Fast
GLM-4.6V-FlashZhipu AI10B7.0 GBRuns well~53 tok/s282,562Fast
Fara1.5-9BMicrosoft9.4B7.0 GBRuns well~55 tok/s14,526Fast
Olmo-3-7B-InstructAi27.3B7.2 GBRuns well~73 tok/s406,776Fast
gemma-4-12B-itGoogle DeepMind12B9.4 GBRuns well~46 tok/s2,963,990Fast
Ministral-3-14B-Instruct-2512Mistral AI14B9.7 GBRuns well~40 tok/s198,169Fast
Nemotron-Labs-Diffusion-14BNVIDIA14B9.7 GBRuns well~40 tok/s6,554Fast
Cosmos3-NanoNVIDIA16B11 GBTight~34 tok/s313,164Fast

Sorted by memory, smallest first — click any column to change it.

Upgrade

Including ERNIE-4.5-21B-A3B-PT (22B) — around ~253 tok/s on a typical machine that size.

See everything 16GB VRAM runs →

All figures at Q4_K_M quantization, 4K context. “Memory” includes the model, its context cache and runtime overhead. Speed is calculated, not benchmarked: 504 GB/s × 0.65 ÷ bytes read per token, assuming a RTX 4070. Scale it by your own machine's bandwidth from the list above. Mixture-of-experts models read only their active parameters per token, which is why some large ones outrun smaller dense models.

Image generators that also fit

These make pictures rather than text. Their memory use scales with the resolution you generate at, not conversation length, so treat these figures as a floor rather than a ceiling.

What you'd need more memory for

The next models up, and what they ask for.

Other hardware

Verified against HuggingFace on 2026-08-08New to this? Start here →Runs on a laptop and up · last 12 months only