Chutes AI

Chutes is a serverless AI compute platform built on decentralized, Bittensor-integrated infrastructure, giving developers immediate access to the latest open-source models (Qwen, DeepSeek, Gemma, Mistral, GLM, Kimi) across text, image, video, speech, and music as soon as they're released. It offers always-on infrastructure with no cold starts, trusted-execution-environment (TEE) options for private computation, and the ability to deploy proprietary/custom models with minimal setup — plus two consumer-facing apps, Chutes Chat and Chutes Search, built on the same infrastructure. Pricing combines subscription tiers (Plus $10/mo, Pro $20/mo, Enterprise custom) with pay-as-you-go token pricing that varies by model ($0.10–$0.45 per million input tokens, $0.28–$15 output, depending on the model). A real consideration: because it's built on decentralized compute rather than a single hyperscaler, evaluate it directly against your latency/reliability requirements rather than assuming hyperscaler-equivalent SLAs. Good fit for a developer who wants day-one access to new open-source models without managing GPU infrastructure themselves.