edrickdch

Just updated the README with a screenshot of the app. appreciate the feedback

Works on Sonoma 14.4 and up. The current implementation is a bit slow on large codebases

I built Promptor after realizing I was paying US $0.30 per o3 request in Cursor.

Coding with o3 is amazing, but with ChatGPT Plus we already get 100 o3 requests every week…

so why not copy the entire repo into the chat?

clearly there's a cost arbitrage here in planning the feature or fix with o3 from the ChatGPT Web UI. Then, using gemini 2.5 or sonnet 3.7 to implement it.

Why I built Blooming

I kept bouncing between model-specific UIs (Kling 1.6, Hailuo Minimax, Midjourney, …) and lost hours going through prompts, AI images and videos. I wanted a single canvas that lets me:

1. Sketch an idea visually – like wiring Lego blocks

2. Swap any model in seconds – no tab switching, see all the assets

3. Download the output easily – no watermarks

You can easily iterate on the same prompts, compare image model outputs, compare video model outputs, turn images into videos.

How it works under the hood

- Frontend: Next.js 15 + React-Flow for node graph

- Backend: Next.js + Python scripts

- Inference: Models from Replicate and OpenRouter

Roadmap

- Add more models

- Export assets into an editor

- Upload your own assets

- Websockets to enable multiple artists to collaborate

Feedback wanted:

- Any bottlenecks?

- Which model/node is highest on your wish-list?

Happy to answer anything! Would love to know how you’d use this, if you’d want to see anything different