Runware
Runware — cheap image generation via a single synchronous call. Image-only — no video. A media provider, wired with createRunwareMediaAdapter.
Runware is a media provider for image generation — often the cheapest
route for FLUX and similar models. Like fal, it's wired with an adapter, not
createOpenAICompatible.
Image only. Runware has no video path in
ai-lcr. Use fal or Kunavo formodality: "video"routes.
import { createRunwareMediaAdapter } from "ai-lcr";
const runware = createRunwareMediaAdapter({
apiKey: process.env.RUNWARE_API_KEY!,
});Unlike fal's queue, Runware returns the image synchronously from a single
call — no polling. Model ids are Runware's vendor:NN@N style, used as the
route's externalId: runware:100@1 (FLUX Schnell), google:4@3 (Nano Banana
2). Runware bills per image / per compute, so the price comes from the pricing
you declare on each route.
registry: {
"bfl/flux-schnell": {
id: "bfl/flux-schnell",
modality: "image",
routes: [
{ provider: "runware", externalId: "runware:100@1", pricing: { unit: "image", cents: 0.14 } },
],
},
}See Image & video for the full media flow and the bundled model catalog.
fal
fal — image and video generation via one async queue API. A media provider, so you wire it with createFalMediaAdapter, not createOpenAICompatible.
Overview
lcr("...") returns a standard AI SDK model, so anything that accepts one works unchanged — the Vercel AI SDK itself, Mastra agents, and any framework built on the AI SDK.