◇ Vision API · Beta · Free

Fast, cheap image analysis.
Drop-in for Gemini Flash.

CPU-only vision pipeline — CLIP concepts, YOLO objects, OCR, face emotion, all in a single 900 ms call. Stateless, horizontally scalable, beta-free while you kick the tyres.

Get a free API key →Read the docs
01

900 ms p50

Full pipeline runs on a 4-vCPU node. No GPU required, no warm-up latency on warm workers.

02

Two API shapes

Retina-native gives you concepts/objects/OCR/faces structured. Gemini-compat is a one-line swap.

03

Free during beta

50 images per key per day. No credit card. Real production use, not a demo sandbox.

◇ Drop-in usage

Swap one line of code.

If you're already using the Gemini SDK, point the API endpoint at retina.frank.ink and keep the same request/response shape.

from google import generativeai as genai

genai.configure(
    api_key="rk_live_...",
    transport="rest",
    client_options={
        "api_endpoint": "retina.frank.ink",
    },
)
model = genai.GenerativeModel("gemini-flash-latest")
result = model.generate_content(
    ["Describe this image", img]
)
print(result.text)