Models
Find a model, compare its price and limits, and copy a setup for your tool.
Models is public. Browse it without signing in, then use a model's full id in your tool or API request.
Find a model for your request
Search by name, id or maker. Choose a maker in the rail, or open More providers for the rest. Choose Reasoning, Tools, Vision, Images or Embeddings in the rail. In a narrow window, use the Capabilities menu for the same filters.
The page opens on a directory of cards. A card shows one price line and the context window. All models is a list with Model, Context, Input / 1M and Output / 1M columns. Sort by Featured, Newest, Cheapest or Longest context on that list. Popular is available when Deference has enough usage to rank models.
Open a card to check the model in more detail. On the list, each row has Copy model ID. Without search or filters, the list starts with 40 rows; Show all {count} displays the rest.
Filters stay in the URL. / shows OpenAI embedding models. See Choose a model for endpoint selection and all capability filters.
Check the price and limits
On a model's page, check its input and output prices, context window and supported capabilities. OpenRouter's price plus OpenRouter's 5% platform fee. Deference adds no markup. See Pricing.
Use the limits published for the model when a coding tool asks for them. A model must support the capability your request needs, such as vision or tool calling.
Connect your tool
Use Copy model ID for the exact id, or choose a tool under Set up in to copy settings with that id filled in. The page includes setups for Claude Code, Codex, Cursor, OpenCode, curl and Python.
For a complete walkthrough, open the relevant coding tool guide. To send requests from your own code, see API overview.
Press / to search, Escape to clear the search, and J or K to move between rows.