Q01
●Live inference playground
Ask the image.
See the evidence.
Upload one image, compose typed questions, and run the pinned Laya Vision model on the local RTX 3090. Every result below comes from the live API.
Checking GPU gateway
Kept in memory unless you explicitly remember it for this browser tab.
LOCAL SETUP
backend/copy_api_token.ps101
Image input
JPEG, PNG, or WebP · max 3 MB decoded · max 16 MP
＋Drop an image hereor choose a file from your computer
02
Questions
Start with an editable preset or build a small batch.
03 / INFERENCE
Requests are processed one at a time by the local GPU.
Model weights are CC BY-NC-SA 4.0. Results may be wrong or overconfident. The local host has no uptime SLA.