PROVIDERS / LIST YOUR INFERENCE

List your inference on ModelIndex

ModelIndex compares every provider serving a model on price, measured speed and capabilities. Most listings come from OpenRouter; if you sell inference directly, submit it here. A person checks every price against your pricing page before anything goes live. Listing is free.

Listing criteria

  1. You serve a model already on ModelIndex. Listings attach to models in the index. Models we don't track yet can't be listed until they're added.
  2. Your prices are machine-readable. A public JSON file, CSV file or API endpoint we can fetch automatically, listing each model with USD per 1M input and output tokens. We keep your listing in step with it, so there's nothing to resubmit when prices change.
  3. Your prices are public. A human-readable pricing page that matches the machine-readable source. No negotiated, invite-only or contact-sales pricing.
  4. Anyone can sign up. Self-serve account and API key, with no sales call required.
  5. Your API is documented. Public API docs and the model id you serve each model under.
  6. You disclose how you serve it. Quantization, context length, max output, data retention, and whether you resell capacity from other sellers. Resold (marketplace) capacity is listed with a MARKETPLACE label and never shown as best price.
  7. Your source stays reachable. If it stops responding or stops matching your pricing page, your listing is marked stale and excluded from best price until it's fixed.

How review works

We fetch your pricing source, check it against your pricing page, and match your models to the ones in the index. Your endpoints then appear with a DIRECT label that links to your pricing page, and compete for best price like any other provider (marketplace listings excepted). We never ask for API keys through this form; if your source needs one, or you offer a test key for speed measurements, we'll follow up by email.

Pricing feed format

Publishing in this shape means we only need to match your model ids to ours. Prices are USD per 1M tokens; everything except id, input and output is optional. We also read an OpenAI-compatible /v1/models that includes per-token pricing (OpenRouter style). For anything else, describe it in the form.

{
  "models": [
    {
      "id": "glm-5.3-flash",
      "input": 0.05,
      "output": 0.20,
      "cache_read": 0.01,
      "context_length": 200000,
      "max_output": 65536,
      "quantization": "fp8",
      "variant": "Zero data retention"
    }
  ]
}
id,input,output,cache_read,context_length,max_output,quantization,variant
glm-5.3-flash,0.05,0.20,0.01,200000,65536,fp8,
glm-5.3-flash,0.06,0.24,,200000,65536,fp8,Zero data retention
Your company
Pricing

No need to type in your models. Point us at a source we can fetch automatically, listing each model you serve with USD per 1M input and output tokens, and we'll keep your listing in step with it.

How you serve models
Anything else