ManifoldGen
Pay-as-you-go generative media APIs: images, video, music, speech, and editing on one prepaid balance, with a hosted MCP server for AI agents.
This site serves clean markdown for AI agents and assistants. Every page exists as HTML at its canonical URL and as raw markdown by appending .md. Prices quoted across these pages are machine-readable at https://manifoldgen.com/api/pricing.
Agents can act on this platform directly through the hosted MCP endpoint https://manifoldgen.com/api/mcp (tools: get_pricing, search_media, generate_media, get_job, list_jobs) or the REST API with an sk-mg- key.
Pages
- AI image generation API from $0.04 per image with frontier models: ManifoldGen exposes image generation, editing, outpainting, relighting, and upscaling through one API. Base generations cost $0.04, frontier models including GPT Image 2, Nano Banana 2, Grok Imagine, and FLUX.2 are selectable per call, and edits are priced flat.
- Pay-as-you-go AI video generation API with no subscription: ManifoldGen generates AI video through a REST API and MCP server on prepaid credits. Videos average $1.71 settled from measured GPU time, jobs are durable and resumable, output is AV1-encoded, and there is no subscription required.
- MCP server that generates images, video, and music from your AI agent: ManifoldGen hosts a Streamable HTTP MCP endpoint at manifoldgen.com/api/mcp exposing semantic media search, pricing, credit-backed generation, and durable job retrieval. Claude, Codex, or any MCP client can generate media directly with a prepaid sk-mg- API key.
- AI music generation API with full commercial tracks from $0.35: ManifoldGen generates complete songs and sound effects through a simple API. Music tracks from 30 to 300 seconds cost $0.35 by default, sound effects settle from measured compute, and finished audio is searchable and retrievable by job ID.
- One API for images, video, music, and voice instead of five subscriptions: ManifoldGen consolidates image generation, video, music, sound effects, speech, transcription, and editing behind one API key and prepaid balance — replacing stacks like Midjourney + CapCut + ElevenLabs + separate matting tools with published per-asset pricing.
- Semantic search over hundreds of thousands of AI-generated images and videos: ManifoldGen's public gallery is searchable by meaning — describe what you want and get matching generated images, videos, and audio with CDN URLs. Free to query via REST or MCP, no key required for public assets.
- Text-to-speech API at $0.005 per 100 characters with transcription: ManifoldGen serves speech synthesis and transcription through one API. Speech costs $0.005 per 100 characters, transcription $0.02 per minute, both settled against prepaid credits alongside image, video, and music generation.
- Train custom LoRA models through an API for consistent characters and styles: ManifoldGen provides LoRA training through REST endpoints — upload a dataset, start training, receive a reusable adapter that keeps characters and styles consistent across generations. Training and inference share one prepaid-credit account.
- Video background removal API producing transparent WebM from $0.10 per second: ManifoldGen removes backgrounds from video via API, returning durable transparent VP9/WebM files. Priced at $0.10 per second of source video up to 30 seconds, with circuit-breaker failover to a standby GPU queue so submissions do not silently drop.
- Video-to-video restyle and character animation APIs with fixed pre-dispatch pricing: ManifoldGen runs Wan 2.2 video restyle from $0.48 per default clip and character animation at $0.75 fixed for a five-second standard clip — priced before dispatch, settled on prepaid credits, delivered as durable files.