Overview

APIMart provides a single OpenAI‑compatible endpoint that lets developers call more than 190 AI models from a single key. The gateway handles chat, image, video, and audio models and charges only for requests that return a response. If you already use an OpenAI SDK, integration is a simple parameter change, preserving your current code and avoiding duplicate key management.

The platform pulls models from major vendors, OpenAI, Anthropic, Google, ByteDance, Qwen, and Kimi, and routes calls based on model name. You can request a GPT‑3.5 model, a Claude‑Sonnet 4.5 prompt, or a video generation model like Veo 3.1‑fast using the same URL (https://api.apimart.ai/v1). Each call is logged in the APIMart console where you can see request counts, total tokens, spend trends, and per‑model distribution. The console also shows the current balance in credits and historical consumption, making it easy to track allocated budgets without juggling separate vendor dashboards.

Video creation works by posting a JSON body that specifies the model (e.g., “kling_v3”), a textual prompt, and a duration (e.g., 5 seconds). The gateway forwards this request to the chosen video model provider, receives the generated clip, and returns it as a binary stream or a URL pointing to the media file. The same pattern applies to image generation: you send a prompt and optional reference images, and the response contains the image data or a link. The system supports reference‑based editing and multi‑turn prompts for image and text models, with resolutions up to 4K for many providers.

APIMart suits teams that need to experiment with several models without managing separate API keys or billing accounts. It is especially useful for developers building AI features that may shift between chat, image, or video capabilities, or for operations teams that want a single place to monitor usage and troubleshooting. It may not be ideal for users who require granular control over provider‑specific pricing nuances or for those who need real‑time streaming endpoints for large video files, as the documentation does not detail streaming behavior or bandwidth limits.

The site offers a console with request distribution charts, credit balances, and token usage per period, but it does not publish explicit pricing tables or rate limits. Detailed SDK or CLI examples are not included beyond the generic OpenAI‑compatible call structure.