SpicyAPI
SpicyAPI is an uncensored AI platform that provides access to 171 AI models including 128 uncensored editions, with no platform-level content classifier or approval queue. It features a no-code browser Studio for creating images, videos, short films, and editing clips, alongside a free built-in video editor for trimming, adding text, and music. The platform also offers an OpenAI-compatible API for developers, supporting text-to-image, text-to-video, image-to-video, face swap, image editing, and chat generation workflows.
Core Capabilities
The full model catalog includes 62 video models, 67 image models, 26 text models, and 16 audio models, with dedicated Spicy LoRA-tuned editions of leading models optimized for unrestricted generation. The no-code Studio runs entirely in the browser, supporting end-to-end creation of images, short videos, and voiced short films up to 15 minutes in length, with a canvas workspace for arranging multiple shots and takes. The built-in free video editor supports trimming, joining clips, adding text overlays, and attaching background music without requiring external tools.
The unified OpenAI-compatible API works across all video, image, and chat models using a single API key, with support for standard OpenAI streaming for chat endpoints and a dedicated /v1/videos endpoint for video generation tasks. Additional platform features include public leaderboards, a blind head-to-head model arena, and hands-on user reviews to compare real-world model performance, plus full LoRA support for running custom fine-tunes on compatible image and video models.
SpicyAPI integrates natively with common creative and automation tools including ComfyUI, n8n, SillyTavern, and popular roleplay applications. For AI agents and coding tools, the platform provides a dedicated MCP server, multiple official SDKs for TypeScript, Python, Go, PHP, and Java, and a CLI for scripted generation workflows. Users can set custom retention periods for three separate categories of stored data: generated media, prompt text, and uploaded reference files, with manual deletion available for any completed generation at any time.
Intended Audiences
This platform is built for creators needing uncensored AI content for images, videos, and short films who want a no-code workflow without platform content restrictions. It supports developers building applications with an OpenAI-compatible uncensored AI API, including roleplay app developers needing access to uncensored LLMs and media generation models. It is also designed for AI agents and coding tools that require direct access to unrestricted model catalogs, and users building automation workflows using tools like n8n and ComfyUI.
Pricing and Workflow
SpicyAPI uses a pay-as-you-go model with no required subscriptions, with all pricing denominated in US dollars per unit: per generated image, per second of generated video, or per million tokens for text models. Users can top up a single shared US-dollar balance via card or crypto that works across both the no-code Studio and the developer API. The in-browser Playground displays a full cost estimate for any generation before the user submits the request, so there are no unexpected charges.
Failed or timed-out tasks are generally not charged, though some individual models apply charges for content refusals, with these terms clearly listed on each model's dedicated page. The platform reserves the estimated cost when a task is created, then settles the final charge based on actual usage once the task completes, releasing any unused portion of the reserved balance back to the user's account. Idempotency keys are supported to prevent duplicate charges if a network error requires resubmitting a request.
Supported Limitations
While SpicyAPI runs no platform-level content classifiers, some mainstream third-party models included in the catalog retain their own native content filters, and these restrictions are clearly noted on each individual model page. Stored data is kept for a limited time by default: generated media for about 14 days, prompt text for up to 30 days and uploaded reference files for 1 day, so users should download anything they want to keep before it expires. All legal and consent responsibilities for generated content remain fully with the customer, as the platform does not moderate or review user-submitted prompts or outputs.
Comments