# ClockingCloud

> ClockingCloud is a GPU cloud built only for generative media. It provisions GPUs and serverless inference endpoints tuned for image and video generation with open models, billed per second from prepaid credit.

- Install the CLI: `brew install clockingcloud`
- API base URL: https://clockingcloud.com/api/v1 (OpenAPI spec in the docs)
- Billing: prepaid credit, per-second metering. Compute stops automatically before the balance runs out.
- Inference endpoints use their own API key, separate from the account key, and can be rotated.
- Generated images and videos are stored privately and returned as short-lived signed links.

## Docs
- [Documentation home](https://clockingcloud.com/docs): Getting started, CLI, API and guides
- [Quickstart](https://clockingcloud.com/docs/quickstart)
- [Installation](https://clockingcloud.com/docs/installation)
- [Authentication](https://clockingcloud.com/docs/authentication)
- [Concepts](https://clockingcloud.com/docs/concepts)
- [Running models](https://clockingcloud.com/docs/running-models)
- [Deploying instances](https://clockingcloud.com/docs/deploying-instances)
- [Storage](https://clockingcloud.com/docs/storage)
- [SSH access](https://clockingcloud.com/docs/ssh)
- [Observability](https://clockingcloud.com/docs/observability)
- [Billing](https://clockingcloud.com/docs/billing)
- [CLI reference](https://clockingcloud.com/docs/cli)
- [Configuration reference](https://clockingcloud.com/docs/configuration)
- [REST API](https://clockingcloud.com/docs/api)
- [OpenAPI spec](https://clockingcloud.com/docs/openapi)
- [Models](https://clockingcloud.com/docs/models)
- [GPUs](https://clockingcloud.com/docs/gpus)
- [Errors](https://clockingcloud.com/docs/errors)
- [Limits](https://clockingcloud.com/docs/limits)
- [Security](https://clockingcloud.com/docs/security)
- [Troubleshooting](https://clockingcloud.com/docs/troubleshooting)
- [FAQ](https://clockingcloud.com/docs/faq)
- [Changelog](https://clockingcloud.com/docs/changelog)

## Models
- Z-Image Turbo (Image, by Tongyi-MAI): Fast, distilled text-to-image with strong photorealism and text rendering. About 2 seconds per image. Minimum 24 GB VRAM. License: Apache 2.0. Model card: https://huggingface.co/Tongyi-MAI/Z-Image-Turbo
- FLUX.1 schnell (Image, by Black Forest Labs): Four-step text-to-image from the FLUX family. Great all-rounder for product shots and illustration. Minimum 24 GB VRAM. License: Apache 2.0. Model card: https://huggingface.co/black-forest-labs/FLUX.1-schnell
- Wan 2.2 5B text to video (Video, by Wan-AI): Up to 5 second 720p clips at 24 fps from a text prompt. Minimum 24 GB VRAM. License: Apache 2.0. Model card: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B
- Qwen3 8B (Text, by Qwen): Chat, reasoning and tool use in 119 languages. Served with vLLM. Minimum 24 GB VRAM. License: Apache 2.0. Model card: https://huggingface.co/Qwen/Qwen3-8B
- gpt-oss 20B (Text, by OpenAI): OpenAI's open-weight reasoning model for agents and coding. Served with vLLM. Minimum 24 GB VRAM. License: Apache 2.0. Model card: https://huggingface.co/openai/gpt-oss-20b

## Blog
- [Robots that imagine before they move](https://clockingcloud.com/blog/robots-that-imagine-before-they-move): Video generation is becoming the training ground and the inner eye of robots. A plain-English guide to world models, generated training data and why the newest robot brains run at the edge.
- [The best image is already in the noise](https://clockingcloud.com/blog/the-best-image-is-already-in-the-noise): A plain-English guide to inference-time search: the newest way to get sharper images and steadier video out of the model you already use, by spending GPU time on looking instead of guessing.

## Optional
- [Platform status](https://clockingcloud.com/status): Live platform health and GPU launch stability
- [Terms](https://clockingcloud.com/terms)
- [Privacy](https://clockingcloud.com/privacy)