Run AI with deAPI

deAPI is a unified API that gives you instant access to open-source AI models for text-to-image, text-to-speech, video transcription, and more. Instead of setting up GPUs yourself, requests are routed through a decentralized GPU cloud, making AI inference fast, scalable, and cost-efficient.

deAPI supports popular open-source models including LTX Video, Qwen Edit, FLUX.1 Schnell, Z-Image Turbo, Kokoro, Whisper, BGE M3, and Nanonets OCR for various AI tasks. You can switch between models through a single endpoint. The model documentation is always updated with all supported models.

deAPI is built for developers, startups, SaaS companies, and AI builders who want to integrate powerful AI features without managing infrastructure. Whether you're building an app, marketplace, or internal tool, you can use deAPI instantly with one API key.

Sign up at deAPI.ai, claim your free credits, and check the API docs for quickstart examples. Try our Playground for instant testing without code. You can make your first AI call within minutes using curl, Python, or JavaScript.

deAPI complements OpenAI perfectly—you can continue using GPT models for chat and text generation while leveraging deAPI for cost-effective AI: image generation, video creation, text-to-speech, transcription, and more. For example, our Whisper Large v3 transcription uses the same model as OpenAI but at a fraction of the cost. Plus, deAPI accepts direct YouTube links for transcription, eliminating the need to download, store, and upload large video files—saving you on storage costs and transfer fees.

No GPU or special hardware required. deAPI runs entirely on our decentralized GPU infrastructure, so you can access powerful AI models with just an API key. Scale your applications instantly without capital investment in expensive hardware, and pay only for what you use—no maintenance, no setup, no infrastructure costs.

Absolutely. All models available through deAPI are open-source with commercial-friendly licenses, meaning you can integrate them into any commercial project without restrictions or additional licensing fees. deAPI already powers startups, SaaS platforms, and marketplace applications worldwide. Build, launch, and scale your commercial AI features with confidence.

Pricing is simple and usage-based. For example: Text-to-Image starts at $0.00119 per image (512×512), Text-to-Speech at $0.77 per 1M characters, Video-to-Text at $0.021 per hour. Because requests run on real distributed GPUs worldwide, inference can be significantly cheaper than providers like OpenAI or Stability AI. New users get $5 in free credits to test the API instantly.

Yes. deAPI is designed for scalable production workloads. You can serve hundreds of requests per minute with automatic load balancing across GPUs. Startups, SaaS platforms, and marketplaces already use it to power AI features in their apps.

Most generations (text-to-image, speech synthesis, video transcription) complete in seconds, depending on the model and prompt size. The decentralized GPU routing ensures requests are always served by the fastest available hardware.

At the moment, deAPI focuses on providing a reliable unified API for popular open-source models. Custom model upload is on the roadmap. For now, you can choose from many optimized models ready to use out of the box.

deAPI has generous rate limits suitable for most applications. Image generation supports up to 2048×2048 resolution, video/audio transcription handles files up to 80MB, and you can make up to 300 requests per minute after any payments. Higher limits are available for production apps — just contact support.

deAPI uses a RESTful API with JSON requests and responses. Authentication is done via API key in the header. Integration typically takes less than 10 minutes — just send a POST request with your prompt and model selection.

Your prompts and generated content are not stored permanently. They may be retained for a short period after processing for operational and reliability purposes, then deleted. You receive the rights to use the generated content under the license of the model you select. Please review the license terms of each model before use. You are responsible for ensuring that your prompts and generated content do not infringe on third-party rights, such as trademarks or copyrighted material. We do not use your data for training and we do not share it with third parties.