P.S. These are just me spinning up a few Cloudflare Pages for the Bolt.diy stable and main branches and adding a Custom Domain. Takes ~5 minutes and is a nice way to deploy it. It can even be easily automated to stay up to date with a GitHub action.
You can do the same for free. It’s just following my tutorial on Deploying Bolt.diy with Cloudflare Pages (the easy way!). Cloudflare only works with static pages so the build step optimizes and “compiles” Bolt.diy to run on it. It’s way lighter and all tasks just runs client side in the Web Container.
Downside is that it will not connect to localhost, if you wanted to connect to Ollama or LMStudio, you would need to port forward and host it through HTTPS. You can do this with tools like ngrok easily, but I haven’t fully tested it myself (I generally use API’s).
What I have dsicovered is that if I import a project from Bolt.New and try using a LLM I get the same generic error.
However IF I start a new project in Bolt.DIY the LLM’s work fine.
Any ideas how I can get the LLM’s to work for imported projects ?
Oh, that was it. @d.lynch made me realize that I hadn’t tried using a NEW project. I was only trying to continue a previous project I had started in bolt.new. So I started a new one, and now OpenAI has no issues at all. So the issue is only with existing projects.
Try to delete the .bolt folder before importing. Dont know if it helps but I can remember it helped one time for me.
If this does not help, maybe your project is to big to work with OpenAI? I´ve got a project that is also not working with OpenAI because it has to many files/tokens/…, but works fine with Google Gemini 2.0 Flash.
(just preparing a youtube video/showcase of it, its the task-list-advanced I show in my first video)
Damn it must be something with my API keys I guess? I locally hosted same issue, cloudflare hosted same issue. I tried your host and welp nothing! Still the same I don’t want to be stuck with Cursor
If I start a brand new application via bolt.diy, I have no problem. But as soon as I import an existing one (albeit previously created via bolt.diy) from a folder or GitHub, I get the same errors and can’t continue working with the AI.
I’m having the same issue. To expand on some of the troubleshooting that’s been mentioned, I’ve tried the following:
Double checked my credit balances
Renamed .env to old.env. Use the API key in the UI
Created a brand new API key and used it in the UI
Spun up a new project using the “To Do List” example project on the front page. This worked, no problem
Exported my project and imported it as a new project
Turned everything off. Restrated my pc.
Tried different browsers. Main browser for bolt is Canary
I tried the latest build as well and had issues there which is when I went back to 0.0.3.
I reinstalled 0.0.3 this morning before testing all this and it was working no problem all day, then all of a sudden it stopped, and for the life of me, I can’t get it up and running again.
I dont use GPT-4o … I use Google Gemini 2.0 Flash as you seen in the screenshot.
Tried OpenAi before I switched to google, because of these limitations. Think it also has some more limitations based on the “Tier” you are on which depends on how much credit you spent.