modlens/relay

Hosted vision relay — OpenAI-compatible /v1

Your text-only model
can see.

One endpoint, one key. Point ModLens, or any OpenAI-compatible client, at the relay and every screenshot, chart, and scan comes back as structured evidence your model can read.

Try the playground Get your key →
point the CLI at the relay 
# three lines, then every image just works
$ modlens config set openai.baseUrl https://relay.modlens.dev/v1
$ modlens config set openai.apiKey ml_************
$ modlens config set provider openai

$ modlens -i screenshot.png
{ "provider": "openai", "result": { "description": "…",
    "text_content": "…", "regions": […] } }

FIG. 01 — RELAY PATH: CLIENT → /v1/CHAT/COMPLETIONS → VISION MODEL

The CLI

BYO key. Free forever.

ModLens is an open-source CLI. Bring your own Gemini, OpenAI-compatible, or Claude key and it costs you nothing, today and always. No account, no meter, no catch.

The relay

For when you don't want to manage keys.

No provider signup, no quota juggling, no config drift across machines. Sign in, copy one key, done. The relay sells convenience, not capability.

How it works

Three steps to sight

  1. 01

    Sign in with GitHub

    One click. No forms, no credit card for the free tier.

  2. 02

    Grab your key

    A single ml_ key with a daily free allowance, ready immediately.

  3. 03

    Point your client at the endpoint

    ModLens, or any OpenAI-compatible client, talks to /v1/chat/completions. Nothing else to change.

FIG. 02 — ONBOARDING SEQUENCE, LEFT TO RIGHT

Pricing

Two dials, no fine print

Free

$0/mo

  • Small daily credit allowance
  • Full playground access
  • Same endpoint, same JSON
Start free

Pro

$X/mo

  • Monthly credit quota
  • Higher rate limits
  • Usage dashboard
Go Pro

Credits-based metering: each request costs credits, heavier models cost more per request. Your dashboard shows the exact burn.

FIG. 03 — PRICE SHEET, SUBJECT TO CALIBRATION