ToolsCompaniesAbout

Runware API

by Runware@runwareSan Francisco, California, United StatesNo ratings yet

Model APIs, serverless GPUs and reserved compute for AI workloads

Visit runware.ai
Tool type
Model API and serverless GPU platform
Made for
Developers, product teams and enterprises building AI features
Runs on
Hosted API endpoints, Serverless GPUs and Sonic Pods
Requests go to the hosted REST and WebSocket API; Serverless workloads run on Runware GPU workers; dedicated clusters run on Sonic Pods; the MCP server is hosted and needs no local install.
Last checked
29 Oct 2026
6 product sources

From @runware, about Runware API

Recent important posts · launches, features, availability

Open on X

What it does

Runware takes API requests containing one or more tasks, each naming a model, and returns generated outputs such as images, video, audio, language model responses and 3D assets. Tasks can be sent over REST, WebSockets, or with a webhook per task, and several modalities can be batched in one call. Developers can also upload their own container or model code to its Serverless GPU platform, where it runs on managed GPU workers that scale with demand.

People use it to call hosted models from their own applications, to train style LoRAs and generate with them through the same API, or to deploy custom workloads without managing GPU servers. The MCP server lets an agent such as Claude or Cursor reach the model catalog through one connection. Its distinguishing claims are a very large model library, per-second GPU billing and reserved capacity options.

It suits developers and product teams building AI features who want one API and pay-as-you-go billing. Access requires a Runware account and an API key; Serverless access is granted after a review of fit and capacity. Pricing varies by model and GPU, reserved and bare-metal options involve commitments or sales contact, and some model prices and capacity terms are subject to change.

What it can do

  • Generate images, video, audio, language and 3D outputs through one API endpoint with a single key and billing.
  • Batch different modalities in a single request, or use REST, WebSockets or per-task webhooks to receive results.
  • Run custom models or containers on Serverless GPUs that scale from zero and bill per second while workers are allocated.
  • Train a style LoRA from an uploaded image dataset and use it in normal image requests on the same API.
  • Connect an MCP-compatible agent to a hosted Runware server that exposes the model catalog as tools.
  • Reserve GPU capacity on a term or deploy dedicated bare-metal clusters on Sonic Pods.

Works with

  • Claude
  • Cursor
  • Codex
  • VS Code
  • Claude Code
  • Python SDK (runware)
  • TypeScript SDK (@runware/sdk)

Before you choose

  • Serverless access is granted after Runware reviews fit and capacity, so it is not instant self-serve for every account.
  • GPU rates fluctuate and listed prices are subject to change; reserved capacity and multi-GPU layouts may require contacting the team.
  • Pay-as-you-go workloads draw from a shared GPU pool, so capacity may depend on availability unless reserved.

Listed under

Get started

Sign up, then use the API key with the REST endpoint, SDKs or the hosted MCP server. Serverless requires requesting access from the dashboard.

  1. Create a Runware account at the Get Started page and obtain an API key.
  2. Call the API at https://api.runware.ai/v1 with the Authorization Bearer header, passing an array of tasks that name a model.
  3. For agents, add https://mcp.runware.ai as a custom MCP connector in your client and sign in once with OAuth.
Create an account and get an API key

Plans and prices

Prices in USD

Model API rates vary by model and modality; see the pricing page's model table for per-model rates. Serverless GPU time is billed per second while a worker is allocated, and scheduling wait, image pulls and platform failures are not billed. New business-email sign-ups receive $2 in free credits, which is a one-time allowance rather than a permanent free tier.

Model API (pay as you go)Usage-basedSee the pricing page's model rate table for named model rates; the page directs readers there for exact rates.
  • Per-generation or per-token billing, varies by model
  • New business-email sign-up includes $2 in free credits
Serverless GPUs (pay as you go)Usage-basedExample list rates are shown on the serverless page; GPU rates fluctuate and are subject to change.
  • Per-second billing while workers are allocated
  • Workers scale to zero when idle

Source: runware.ai/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

Ratings and reviews

No ratings for Runware API yet.

Be the first to rate it

What’s changed

Free account

See what’s changed at Runware API

Launches, price moves and pivots, each with a date and a source. Free with an account.

Signals

Free account

See Runware API’s signals

Launch traction, hiring and release activity, tracked over time. Free with an account.

Data and security

Free account

See how Runware API handles your data

Training on your files, compliance, hosting and the models behind it. Free with an account.

Compare with similar tools

Free account

Compare Runware API side by side

Prices, platforms and features against similar tools, with sources. Free with an account.

Where this comes from

How we research
Company website
runware.ai: product, pricing and policy pages
YC directory mirror
Batch and company name
Social accounts
Accounts the company links, and its posts on X
Human review
Checked by a person
knowledge_base
manual_review

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.