Price API and price badge

Two addresses give you UberLLM's current prices without an API key or an account:

• Prices as JSON: GET https://api.uberllm.dev/api/v1/public/prices
• A price badge: https://api.uberllm.dev/api/v1/public/badge/<model-id>.svg

Both show the same price as the model's page on uberllm.dev: the lowest seller price right now. Prices are re-read every 30 seconds.

Prices as JSON

curl "https://api.uberllm.dev/api/v1/public/prices?model=deepseek/deepseek-v4-flash"

{
  "updated_at": "2026-11-13T06:40:12Z",
  "models": [
    {
      "id": "deepseek/deepseek-v4-flash",
      "name": "DeepSeek V4 Flash",
      "type": "text",
      "input_per_m": 0.0091,
      "output_per_m": 0.0182,
      "list_input_per_m": 0.09,
      "list_output_per_m": 0.18,
      "discount_pct": 89.9,
      "unit": "1M tokens",
      "price_text": "$0.0091/M in · $0.0182/M out",
      "page_url": "https://uberllm.dev/models/deepseek/deepseek-v4-flash"
    }
  ]
}

The numbers above were read on 2026-11-13. Yours will differ, because prices move.

What you can ask for

All of these are optional. With none, you get every model you can call right now.

If model names a model that does not exist, or one that nobody serves right now, models is an empty list and the answer is still 200. An old model id that has been renamed answers with the model it became.

What comes back

All prices are in US dollars. A field with no value is left out.

A request can cost more than the price shown when the cheapest sellers are busy. See Pricing for how a request is billed.

From a web page

Any website can read the prices straight from the browser:

const res = await fetch("https://api.uberllm.dev/api/v1/public/prices?type=text&limit=10");
const { models } = await res.json();
for (const m of models) console.log(m.name, m.price_text);

Price badge

A small picture with the model's name and its price now. It updates by itself.

Markdown, for a README:

[![DeepSeek V4 Flash price on UberLLM](https://api.uberllm.dev/api/v1/public/badge/deepseek/deepseek-v4-flash.svg)](https://uberllm.dev/models/deepseek/deepseek-v4-flash?utm_source=embed&utm_medium=referral&utm_campaign=price-badge)

HTML:

<a href="https://uberllm.dev/models/deepseek/deepseek-v4-flash?utm_source=embed&amp;utm_medium=referral&amp;utm_campaign=price-badge">
  <img src="https://api.uberllm.dev/api/v1/public/badge/deepseek/deepseek-v4-flash.svg"
       alt="DeepSeek V4 Flash price on UberLLM" height="20">
</a>

• Put the model id in the address as it is, slashes included, and end it with .svg.
• The badge also says how far the price is below the list price. Add ?discount=0 to show the price alone.
• If nobody serves the model right now, the badge says "not available right now" and shows no price. It goes back to the price when a seller returns.
• A model id that does not exist answers 404.
• The link carries utm_source=embed, so we can tell that a visit came from a badge. It records nothing about your readers. Take it off if you prefer.

Limits and caching

• 300 requests a minute from one IP address on the JSON address. Over that you get 429 with a Retry-After header.
• Badges have no request limit, and a badge for a real model is always a picture: if prices cannot be read for a moment it says "not available right now" instead of breaking.
• JSON answers may be cached for 60 seconds, badges for 5 minutes. Sites that proxy images (GitHub does) can hold a badge a little longer.
• Models made for adult content are left out of both. They are in GET /v1/models, marked "nsfw": true.

Errors

Related

• GET /v1/models: the full model list in the OpenAI format, with context sizes and what each model supports.
• MCP server: the same prices as tools for an AI assistant.
• OpenAPI spec: both addresses are in it.