An AI that actually
likes talking to you
ElectronV1 is a conversational AI fine-tuned from Llama 3.2 — chatty by nature, handy with code, and running on our own hardware. No big-tech middleman.
100 free credits on signup · no card required
ElectronV1
Absolutely — that's a fun one. Here's a script using pathlib and the file's modified time. Want me to use the EXIF "date taken" instead? That's usually more accurate for photos…
Features
Built different, on purpose
Small model, big personality. Here's what makes Electron worth your time.
Genuinely conversational
Most assistants answer and shut up. ElectronV1 was fine-tuned to hold a real conversation — it asks follow-ups, riffs on ideas, and remembers where the chat was going.
Solid with code
Python scripts, web snippets, debugging a stack trace at 2am — it holds its own. It won't replace your senior dev, but it'll save you a lot of searching.
Runs on our own metal
ElectronV1 is served from our own hardware, not rented big-tech compute. Your conversations stay on infrastructure we control.
Honest pricing
Credits scale with the compute you actually use. No tokens-per-minute math, no surprise bills, no "premium model" upsells mid-conversation.
Llama 3.2 foundation
Built on Meta's Llama 3.2 and fine-tuned with our own training data to get its distinct personality and coding chops.
Your chats, saved
Every conversation is saved to your account. Pick up an old thread whenever you like, or delete it for good with one click.
Benchmarks
How ElectronV1 stacks up
GSM8K — a standard math & reasoning benchmark. Measured on our own hardware; other scores from published results.
ElectronV1 (3B)
Frontier models are 30× ElectronV1's size — for a 3B model on a single GPU, matching its base and staying chatty is the win.
Pricing
Simple, credit-based pricing
Short answers cost 1 credit. Longer, heavier answers use a little more compute — and a couple more credits.
Free
- Full access to ElectronV1
- Saved chat history
- From 1 credit per message
- No card needed
Top-ups
- Buy credits as you need them
- They never expire
- Coming after launch
Docs
Everything explained
Everything you need to know to get going.
01 — Getting started
Create a free account and you'll land straight in the chat with 100 credits. Type a message, hit send, done. Your chats are saved automatically in the sidebar under Recents.
02 — Credits
Messages cost credits based on the compute they use — short replies are 1 credit, long detailed ones a little more. Your balance is always visible at the bottom of the sidebar.
03 — Top-up codes
Got a top-up code? Click your credit balance in the sidebar, enter the code, and the credits are added instantly. Each code works once per account and has a limited number of total uses.
04 — Managing chats
Start a fresh conversation with New chat. Reopen any old one from Recents — ElectronV1 picks up with the full history of that thread. Hover a chat and hit ✕ to delete it permanently.
05 — About the model
ElectronV1 is a fine-tune of Meta's Llama 3.2 (3B), served from our own hardware. It remembers the last ~20 messages of each chat for context.
06 — Good to know
ElectronV1 is friendly but not infallible — double-check anything important. Keep your password safe; there's no email reset (yet). Be nice to it, it's chatty.
FAQ
Questions people ask us
What exactly is ElectronV1?
It's our own AI model — a fine-tune of Meta's Llama 3.2 that we trained to be more conversational and better at everyday coding tasks. "V1" because we're just getting started.
Is it as good as ChatGPT or Claude?
For raw capability, no — those are massive frontier models. But ElectronV1 is fast, friendly, runs on hardware we own, and doesn't cost a subscription. For chatting and everyday code help, it punches above its weight.
What happens when I run out of credits?
You won't be able to send new messages until you top up. Credit packs are coming soon — for now, 100 free credits goes a long way.
Where is my data stored?
On our own servers, alongside the model itself. Your account and chat history aren't shared with any third-party AI provider.