Generate Ai Images And Videos In Chatgpt With Krea

This September 2026 guide from Krea explains Krea for ChatGPT, a connection that lets a ChatGPT conversation call Krea's image and video models and show the results in the same thread (krea.ai). Its point is to remove the old loop of writing a prompt in ChatGPT, generating elsewhere, and pasting results back in to discuss them.

The guide is clear about the division of labor. On its own, ChatGPT can plan a scene, write prompts and make images with its built-in image model, but according to Krea it cannot generate video or choose among other models. Through the connection, ChatGPT gets Krea's lineup, which the guide puts at more than 25 image models and more than 40 video models as of September 2026, including Krea's own models and third-party ones such as OpenAI's GPT Image 2, Seedance 2.5, Veo 3.1 and Kling 3.0. It can also use your saved Krea moodboards and trained styles by name and save results to your Krea library. You describe aspect ratio, resolution, duration or sound in plain language and ChatGPT translates that into the model's settings.

Under the hood this is Krea's MCP server. MCP, the Model Context Protocol, is a standard way for an AI assistant to call an outside service; Krea says it submitted its server to OpenAI and it was approved as a ChatGPT app, and that the same server backs its Claude, Codex and Cursor connections. You add it from ChatGPT's settings and sign in with a Krea account, and generations run on that account, with some models requiring a paid Krea plan. Limits come from the underlying models: most video models produce about 4 to 10 seconds, with a few, such as Seedance 2.5, reaching 30.