Compare model features and costs before you build
Get available AI models, token prices, and supported features in one response. Use the catalog to estimate costs and choose which models to test.
Build a shortlist for your workload
First, check the features your app needs. A model may need to read images, call tools, or return JSON. Its context window must also fit the material you send and the response you need.
Next, estimate the cost using both input and output token prices. The balance between the two matters: a short classification response and a long generated report have different usage patterns. The guide below includes a Python example you can adapt to your expected traffic.
Finally, run your own prompts against the models on your shortlist. The catalog tells you what a model supports and costs; the results tell you whether it works for your app. Our leaderboards provide additional measured results, with test methods and run dates.
Try it step by step
Try a request
This request lists the available models. The response below is an example; use the endpoint for current capabilities and prices.
curl https://api.nativeport.ai/inference/v1/models \ -H "Authorization: Bearer $NATIVEPORT_API_KEY"
{
"object": "list",
"data": [{
"id": "openai/gpt-5.4-mini",
"context_window": 400000,
"pricing": {"prompt_usd_per_1m": 0.25, "completion_usd_per_1m": 2.00},
"capabilities": ["chat", "tools", "vision", "json_mode"]
}]
}Compare providers
Compare starting prices and the tasks each provider supports. Features and usage affect the total cost. For measured results, see the benchmarks and how we test.
| Provider | Best suited for | Starting price |
|---|---|---|
| Anthropic | Compare available Claude models and token rates. | metered by Anthropic |
| OpenAI | Compare available GPT models and token rates. | metered by OpenAI |
| Grok | Check the Grok models available for your requests. | metered by Grok (xAI) |
| Hugging Face | Compare supported open models for your workload. | metered by Hugging Face |
More ways to use these APIs
Find models with JSON output support
Check the capabilities list for JSON support before choosing a model. Then use
that model’s supported response-format options and validate the returned data
against your application’s requirements.
NativePort Inference · /inference/v1/models
Compare embedding models, then call one
Check embedding capabilities and prices before choosing the model and endpoint for your search workflow. Test retrieval with your own documents. Changing the model later usually requires embedding the stored documents again, since vectors from different models are not directly comparable.
NativePort Inference · Jina · /inference/v1/models → /jina/embeddings
Find LLM rankings and leaderboards
Use a web search to find published model comparisons, and check what each test measures. NativePort’s own leaderboards include run dates for the capabilities we have tested. Use those results alongside tests on your own prompts.
Serper · /serper/search
Use it with NativePort
Read model details in one request and try your shortlist with the same NativePort key. You can compare providers without funding a separate account for each one.
Before you start
Are these the prices NativePort uses?
The catalog returns each model’s input and output token rates in USD. Model usage is billed at those rates with nothing added per call. Fetch the catalog again when you need current prices.
How do I choose a different model?
Set the model field on the chat-completions request to the model ID you want. Your application controls model selection and any fallback logic.
How often should I refresh the catalog?
Models and prices can change. Use a short-lived cache if you query the list often, and refresh it before a batch that depends on a particular model or price.