Chutes
Serverless, decentralised compute for running open-source AI models in production
Chutes is a serverless AI compute platform that hosts open-source models and rents out GPU time for inference, batch jobs and private workloads. The company describes itself as a decentralised, open-source infrastructure provider handling trillions of tokens per month, with new models added shortly after release and pay-per-use billing.
- Claims to process trillions of tokens per month
- Hosts language, image, video, speech and music models
- OpenAI-compatible chat completions API with streaming
- Per-token and per-second GPU billing, no subscription
- Says new open-source models land within minutes of release
What it is
Chutes runs open-source AI models behind a serverless API, so developers send requests and pay for the tokens and GPU seconds they consume. The company says its platform handles trillions of tokens each month and hosts language, vision, image, video, speech and music models. It targets products that need to scale inference without managing hardware. Access is through an OpenAI-compatible chat completions endpoint with a bearer API key, and responses can be streamed. There is also a consumer chat app called Chutes Chat and a character roleplay product, fictio, built on the platform.
The compute
The platform is described as decentralised and open-source, with models kept permanently warm so requests avoid cold starts. Users can run their own code against any hosted open model, or reserve private GPU capacity billed by the second. Chutes says its team adds newly released open-source models shortly after they appear, and it lists families such as Kimi, GLM, DeepSeek, Qwen, Gemma, MiniMax and Mistral among its hosted options.
Customers and deals
Chutes serves AI developers building products that need inference at scale, plus batch and embedding workloads such as search, recommendation, hate speech detection and content moderation. Pricing is usage-based: a prepaid balance, live per-token rates for each hosted model, and per-second charges for private GPU capacity, with no subscription or minimum commitment.
Its place in AI
The company positions itself as an early home for newly released open-source models, claiming they appear on Chutes within minutes of launch. It also offers consumer-facing apps built on its own infrastructure, including chat and roleplay products.
Questions people ask about Chutes
What does Chutes actually sell?
Serverless access to open-source AI models through an API, plus private GPU capacity for your own code. You pay for the tokens you send and receive, and for GPU time by the second, with no monthly commitment.
How do I call it?
Requests go to an OpenAI-compatible chat completions endpoint using a bearer token from a Chutes API key. Responses can be streamed back line by line, and both text and vision-language models are available.
Which models are hosted?
The company lists families including Kimi, GLM, DeepSeek, Qwen, Gemma, MiniMax and Mistral, along with image, video, speech and music models. It says newly released open-source models are added shortly after launch.
Is there a free tier or contract?
Billing is prepaid and usage-based. You top up a balance and pay only for tokens and GPU time consumed, with live per-token rates published for each model and per-second pricing for private capacity.
Good reading and sources
There is no English Wikipedia article on Chutes yet, so the facts on this page come from the sources listed above: chutes.ai/. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.