Computer Vision
Image Understanding: on-device vs large model
See a photo two ways: a light classifier on your device, and a large vision model with your key.
Production proof: The on-device vs cloud trade-off, made visible. From my edge-vision work: run inference where the data is, and escalate to a large model only when the task needs it. Read the case study →
⚡ Run it live with your own API key no key set
Your key is held in memory for this session (kept for this tab only if you tick the box) and sent only to the provider you choose, directly from your browser. This site has no backend server — you can confirm it in your browser's Network tab: requests go only to the provider, never to this site. Live runs use small, fast models with hard token caps; a typical run costs well under $0.01, billed to your own account.
Honest limits: no website can protect a key from a compromised browser or malicious extension. Use a key with a spending cap, and revoke it after playing if you like.
How it’s engineered
- The light path runs a ~50 MB Vision Transformer classifier entirely in your browser via WebAssembly. It returns ImageNet labels in milliseconds, your photo never leaves the device, and it works offline once cached.
- The large path sends the image to a frontier vision model with your own key, straight from the browser. It does not just label the image, it understands it: scene, relationships and legible text.
- That gap is the real production decision: cheap, private, on-device inference for the common case, escalating to a large model only when the task genuinely needs its depth.