Someone are still asking what Lumine actually is, so let me explain.
Lumine turns creator fees into a shared AI balance that anyone can use. You can try models on the website or connect them to your own app or coding agent. No signup needed.
It also has tools for web, social and onchain data, plus a canvas where you can build and run workflows.
I’ll add more models as the API vault grows. For now, here’s what you can already do with it 🧵
The balance and usage are visible on the Stats page: how much API balance has been earned, how much has been used, what’s left for the day and which models people are using.
Access depends on the shared pool. Up to 35% of the available earned balance funds each UTC day, and requests pause when the daily allowance or balance runs out. The limits are explained on the Rules page.
And yes, the little Lumine on the page does something too. You can drag her around, tap her, or tap four times to open a chat and ask about the site.
I also post dev updates on the homepage, where you can vote on whether an update was useful.
Have a look, try a model, or connect it to something you’re building.
https://t.co/KDy5OutCbN
Someone are still asking what Lumine actually is, so let me explain.
Lumine turns creator fees into a shared AI balance that anyone can use. You can try models on the website or connect them to your own app or coding agent. No signup needed.
It also has tools for web, social and onchain data, plus a canvas where you can build and run workflows.
I’ll add more models as the API vault grows. For now, here’s what you can already do with it 🧵
Agent Canvas is for putting those pieces together.
Describe what you want to do and it creates a workflow you can edit. Or start from scratch, add steps and connect them yourself.
If you’re thinking about building with Orbio but aren’t sure where to start, you can try things out on our canvas first. You can combine text models, decisions, embeddings and tools, then review the calls before running anything.
The tools currently cover:
• X posts scrape, profiles and post lookup
• Instagram and TikTok
• Web search and reading pages
• EVM chain data
You’ll find their arguments, prices and code examples in Connect. Each tool call is capped at 0.1 $CREDIT.
There’s also a request inspector and run history when you need to check what happened.
Decision compare is no longer limited to two models.
Paste one situation per line, pick any models you want, and compare their answers side by side. One model fails? The rest keep running.
Errand's first Bountathon has officially come to an end!
Big congratulations to the 3 winners:
Minimum Viable Agent
Lumine
Caturn
Well done for making it to the podium, rewards will be sent to your addresses shortly.