One API, many production loops.
Pick the one you're building.

Every card below is a real production loop teams ship on deAPI — image, video, speech, music, transcription and agent workflows. One API, one response contract, one decentralized GPU pool under all of them. Pick the loop closest to your product.

Find your loop

Each page shows the API shape, the models in play, and a production-shaped example of what ships.

Image from $0.00141/img

Text-to-Image

Open-weight image catalog behind one call

Route the same prompt to FLUX, Z-image and more. The page covers model tradeoffs, resolution and step controls, and where each checkpoint wins for product shots, art, or UI mocks.

Image ~$0.01/edit

AI Image Editor

Generate, edit, upscale, background-remove

A four-step editing loop on a single credit pool — FLUX-2 Klein to generate, Qwen Image Edit for precise edits, then upscale and cutout. Full pipeline walkthrough inside.

Image

Prompt Enhancer

Model-aware rewrites, not generic cleanup

Each model has its own prompt grammar. Prompt Enhancer rewrites against model-specific guides built from HuggingFace and GitHub sources — for image, img2img, video, and TTS inputs.

Video

Image-to-Video

Animate a still frame into a short clip

Breakdown of duration, motion strength and seed controls across the video models in the catalog — with examples for product shots, social content, and avatar motion.

Video ~$0.04/avatar

AI Video Avatar

Talking-head pipeline in three calls

Chain text-to-image, text-to-speech (Kokoro or Chatterbox) and LTX-2.3 audio-to-video to produce a lip-synced avatar from a description. End-to-end code inside.

Video

Video-to-Text

Whisper large-v3, timestamps, 99 languages

Transcription endpoint that handles video and audio URLs, returns word-level timestamps, and auto-detects language — the page covers chunking strategy and accuracy tradeoffs.

Audio

Text-to-Speech

Kokoro, Chatterbox and more — one endpoint

A single TTS endpoint fronting multiple open-weight voice models. The page covers voice selection, SSML-style controls, and latency characteristics for interactive apps.

Audio

Music Generation

Text-to-music with BPM, key and lyrics control

Generate instrumental or vocal tracks from a description. The page details the parameter surface — genre, BPM, mood, duration, key, lyrics — and where it fits in game, ad, and video pipelines.

Agents & Workflows

n8n AI Workflows

10 nodes — image, video, TTS, transcription

Drop deAPI into n8n as ready-made nodes and wire media generation into any flow without GPU setup. The page lists every node and walks through production automations end-to-end.

Agents & Workflows

Claude Code Skills

Slash commands for media inside the terminal

Eleven open-source skills that add /generate-image, /transcribe, /generate-audio, /generate-video and more to Claude Code. Install steps and skill-by-skill examples on the page.

Agents & Workflows

OpenClaw Agents

Voice, hearing, vision and memory for ClawdBots

Kokoro TTS, Whisper transcription, FLUX / Qwen image generation and BGE-M3 embeddings exposed to an OpenClaw agent through one key. The page shows how each capability plugs in.

Still scoping? Start with $5 of free credits

No credit card. One API across image, video, speech, music and transcription — on a decentralized GPU pool.

Need help wiring a loop? Talk to an engineer.