Models
The live catalog, choosing a model, and what a reply costs.
The catalog
GET /api/runway/models returns free models and the newest paid text models from each supported provider on OpenRouter, with context length, prices per million tokens and availability. Runway refreshes it every ten minutes. Free models are always open; paid models open while Runway holds model credits.
Choosing
The model is part of the signed launch and does not change afterwards. Pick a long context window if people will paste documents, and cheap output tokens if the startup will field lots of short questions. The token market is a separate choice.
Pricing
Free models cost nothing and never touch runway, but OpenRouter caps them with a daily request limit that all of Runway shares. Paid models charge the provider's reported cost for each reply to runway, with no markup. The calculator on the home page shows how many paid replies a balance buys at 1,200 input and 300 output tokens per reply.