What LLM Can I Run? Free GPU & VRAM Model Checker
LLM Model Checker
What LLM can I run on my hardware?
GPU Your graphics card. Auto-detected from your browser, or select manually. RTX 4070
VRAM Video memory available for model weights. Determines which models can run. On Apple Silicon, VRAM is shared with system RAM (unified memory). 12 GB
BW Memory bandwidth in GB/s. Determines inference speed (tokens per second). 504 GB/s
RAM System memory. On Apple Silicon, RAM is unified with VRAM. Model and OS share this pool. 32 GB
Estimates based on browser APIs. Actual specs may vary.
Models you can run
13 models fit your RTX 4070
- 2 Excellent
- 1 Good
- 1 Weak
| Model | Quality | Speed | VRAM | Context |
|---|---|---|---|---|
| Qwen3.5-9B | Excellent | ~71 tok/s | 5GB | 262K ctx |
| Qwen3.5-4B | Excellent | ~176 tok/s | 2GB | 262K ctx |
| Gemma 4 12B | Good | ~59 tok/s | 6GB | 262K ctx |
| GPT-oss 20B | Weak | ~32 tok/s | 11GB | 128K ctx |
See full benchmarks and quality scores for each model.