memory

What can your GPU actually run?

Inspect the graphics adapter and estimate local model fit using weights, runtime overhead, and KV cache—not parameter count alone.

0GPU

0HYBRID

0OOM

qwen3:0.6b

Qwen 3 · 0.6B · text

File0.8 GB

Est. VRAM1.3 GB

CHECK

llama3.2:1b

Llama 3.2 · 1B · text

File1.0 GB

Est. VRAM1.5 GB

CHECK

qwen3.5:0.8b

Qwen 3.5 · 0.8B · text + image

File1.0 GB

Est. VRAM1.5 GB

CHECK

deepseek-r1:1.5b

DeepSeek R1 · 1.5B · text

File1.3 GB

Est. VRAM1.9 GB

CHECK

moondream:1.8bPICK

Moondream 2 · 1.8B · text + image

File1.7 GB

Est. VRAM2.4 GB

CHECK

qwen3-vl:2b

Qwen3-VL · 2B · text + image

File1.9 GB

Est. VRAM2.7 GB

CHECK

qwen2.5-coder:3b

Qwen 2.5 Coder · 3B · text

File1.9 GB

Est. VRAM2.6 GB

CHECK

llama3.2:3b

Llama 3.2 · 3B · text

File2.0 GB

Est. VRAM2.7 GB

CHECK

granite3.2-vision:2bPICK

Granite 3.2 Vision · 2B · text + image

File2.4 GB

Est. VRAM3.2 GB

CHECK

phi4-mini:3.8b

Phi-4 Mini · 3.8B · text

File2.4 GB

Est. VRAM3.2 GB

CHECK

qwen3.5:2b

Qwen 3.5 · 2B · text + image

File2.7 GB

Est. VRAM3.3 GB

CHECK

llava-phi3:3.8b

LLaVA-Phi3 · 3.8B · text + image

File2.9 GB

Est. VRAM3.8 GB

CHECK

qwen2.5vl:3b

Qwen2.5-VL · 3B · text + image

File3.2 GB

Est. VRAM4.1 GB

CHECK

qwen3-vl:4bPICK

Qwen3-VL · 4B · text + image

File3.3 GB

Est. VRAM4.3 GB

CHECK

gemma3:4bPICK

Gemma 3 · 4B · text + image

File3.3 GB

Est. VRAM4.2 GB

CHECK

qwen3.5:4bPICK

Qwen 3.5 · 4B · text + image

File3.4 GB

Est. VRAM4.2 GB

CHECK

mistral:7b

Mistral 7B · 7B · text

File4.1 GB

Est. VRAM5.1 GB

CHECK

llava:7b

LLaVA 1.6 · 7B · text + image

File4.7 GB

Est. VRAM5.8 GB

CHECK

bakllava:7b

BakLLaVA · 7B · text + image

File4.7 GB

Est. VRAM5.8 GB

CHECK

qwen2.5-coder:7bPICK

Qwen 2.5 Coder · 7B · text

File4.7 GB

Est. VRAM5.7 GB

CHECK

llama3.1:8b

Llama 3.1 · 8B · text

File4.7 GB

Est. VRAM5.7 GB

CHECK

qwen3:8b

Qwen 3 · 8B · text

File4.7 GB

Est. VRAM5.7 GB

CHECK

deepseek-r1:8bPICK

DeepSeek R1 · 8B · text

File5.2 GB

Est. VRAM6.2 GB

CHECK

llava-llama3:8b

LLaVA-Llama 3 · 8B · text + image

File5.5 GB

Est. VRAM6.7 GB

CHECK

minicpm-v:8b

MiniCPM-V · 8B · text + image

File5.5 GB

Est. VRAM6.7 GB

CHECK

qwen2.5vl:7b

Qwen2.5-VL · 7B · text + image

File6.0 GB

Est. VRAM7.2 GB

CHECK

qwen3-vl:8b

Qwen3-VL · 8B · text + image

File6.1 GB

Est. VRAM7.4 GB

CHECK

qwen3.5:9b

Qwen 3.5 · 9B · text + image

File6.6 GB

Est. VRAM7.7 GB

CHECK

gemma4:e2b

Gemma 4 Edge · E2B / 5.1B total · text + image

File7.2 GB

Est. VRAM8.3 GB

CHECK

gemma4:12b

Gemma 4 · 12B · text + image

File7.6 GB

Est. VRAM9.0 GB

CHECK

phi4:14b

Phi-4 · 14B · text

File7.7 GB

Est. VRAM9.0 GB

CHECK

llama3.2-vision:11b

Llama 3.2 Vision · 11B · text + image

File7.8 GB

Est. VRAM9.2 GB

CHECK

llava:13b

LLaVA 1.6 · 13B · text + image

File8.0 GB

Est. VRAM9.4 GB

CHECK

gemma3:12b

Gemma 3 · 12B · text + image

File8.1 GB

Est. VRAM9.5 GB

CHECK

qwen2.5-coder:14b

Qwen 2.5 Coder · 14B · text

File9.0 GB

Est. VRAM10.3 GB

CHECK

deepseek-r1:14b

DeepSeek R1 · 14B · text

File9.0 GB

Est. VRAM10.3 GB

CHECK

gemma4:e4b

Gemma 4 Edge · E4B / 8B total · text + image

File9.6 GB

Est. VRAM10.9 GB

CHECK

gpt-oss:20b

GPT-OSS · 20B · text

File11.3 GB

Est. VRAM12.9 GB

CHECK

mistral-small3.1:24b

Mistral Small 3.1 · 24B · text + image

File15.0 GB

Est. VRAM16.6 GB

CHECK

mistral-small3.2:24b

Mistral Small 3.2 · 24B · text + image

File15.0 GB

Est. VRAM16.6 GB

CHECK

qwen3:30b-a3b

Qwen 3 MoE · 30B-A3B · text

File15.9 GB

Est. VRAM17.7 GB

CHECK

qwen2.5-coder:32b

Qwen 2.5 Coder · 32B · text

File16.9 GB

Est. VRAM18.8 GB

CHECK

deepseek-r1:32b

DeepSeek R1 · 32B · text

File16.9 GB

Est. VRAM18.8 GB

CHECK

qwen3.5:27b

Qwen 3.5 · 27B · text + image

File17.0 GB

Est. VRAM18.8 GB

CHECK

gemma3:27b

Gemma 3 · 27B · text + image

File17.0 GB

Est. VRAM18.8 GB

CHECK

gemma4:26b

Gemma 4 MoE · 26B-A4B · text + image

File18.0 GB

Est. VRAM19.8 GB

CHECK

qwen3.5:35b

Qwen 3.5 MoE · 35B-A3B · text + image

File18.4 GB

Est. VRAM20.3 GB

CHECK

qwen3-vl:30b

Qwen3-VL MoE · 30B-A3B · text + image

File20.0 GB

Est. VRAM21.9 GB

CHECK

llava:34b

LLaVA 1.6 · 34B · text + image

File20.0 GB

Est. VRAM22.1 GB

CHECK

qwen3-vl:32b

Qwen3-VL · 32B · text + image

File21.0 GB

Est. VRAM23.0 GB

CHECK

qwen2.5vl:32b

Qwen2.5-VL · 32B · text + image

File21.0 GB

Est. VRAM23.0 GB

CHECK

gemma4:31b

Gemma 4 · 31B · text + image

File21.0 GB

Est. VRAM23.0 GB

CHECK

llama3.3:70b

Llama 3.3 · 70B · text

File36.4 GB

Est. VRAM39.0 GB

CHECK

qwen2.5vl:72b

Qwen2.5-VL · 72B · text + image

File49.0 GB

Est. VRAM51.9 GB

CHECK

llama3.2-vision:90b

Llama 3.2 Vision · 90B · text + image

File55.0 GB

Est. VRAM58.3 GB

CHECK

qwen3.5:122b

Qwen 3.5 MoE · 122B-A10B · text + image

File63.0 GB

Est. VRAM66.5 GB

CHECK

llama4:scout

Llama 4 MoE · 109B-A17B · text + image

File67.0 GB

Est. VRAM70.7 GB

CHECK

qwen3-vl:235b

Qwen3-VL MoE · 235B-A22B · text + image

File143.0 GB

Est. VRAM147.6 GB

CHECK

qwen3.5:397b

Qwen 3.5 MoE · 397B-A17B · text + image

File203.9 GB

Est. VRAM209.6 GB

CHECK

llama4:maverick

Llama 4 MoE · 400B-A17B · text + image

File245.0 GB

Est. VRAM251.2 GB

CHECK

deepseek-r1:671b

DeepSeek R1 MoE · 671B-A37B · text

File344.2 GB

Est. VRAM351.5 GB

CHECK

GPU — GPU: fits in available VRAMHYBRID — HYBRID: requires CPU/RAM offloadOOM — OOM: exceeds the current memory budget

What can your GPU actually run?

War dieses Werkzeug hilfreich?

Verwendung

Sicherheit

Funktionen