I got frustrated paying monthly for something that could run on my Mac, so I built Lucid Voice:
- 100% offline (Nvidia Parakeet + Llama)
- $20 one-time (mainly to cover Apple's notarization costs)
- Runs on surprisingly low-end hardware (M1 base models work fine)
- No cloud, no data collection
Open to feedback on anything - pricing, the tech stack, or if this should just be free: https://lucidvoice.app
Plenty on Reddit saying they did. And I did.
Could the outage be a the result of an "unexpected" surge in account activations / use?
Not much of a "welcome back" ;-)
I spent at least an hour trying to get OpenCode to use a local model and then found a graveyard of PRs begging for Ollama support or even the ability to simply add an OpenAI endpoint in the GUI. I guess the maintainers simply don't care. Tried adding it to the backend config and it kept overwriting/deleting my config. Got frustrated and deleted it. Sorry but not sorry, I shouldn't need another cloud subscription to use your app.
Claude code you can sort of get to work with a bunch of hacks, but it involves setting up a proxy and also isn't supported natively and the tool calling is somewhat messed up.
Warp seemed promising, until I found out the founders would rather alienate their core demographic despite ~900 votes on the GH issue to allow local models https://github.com/warpdotdev/Warp/issues/4339. So I deleted their crappy app, even Cursor provides some basic support for an OpenAI endpoint.
Therefore you can forget completely supplying electricity for an oven, kettle or microwave in most cases.
*Due to local regulations: 800W UK, 600W in some parts of Europe - I believe it's 800W in the US but please DYOR.
Having said all that, their existing PowerStream inverter + battery Power Stations work really well and are literally plug & play. I've been running a setup in the UK and it's drastically reduced our electricity consumption from the grid whilst still being a simple plug & play self-install.
Dead Comment
Back in 2023 one of the cursor devs mentioned [1] that they first convert the HTML to markdown then do n-gram deduplication to remove nav, headers, and footers. The state of the art for chunking has probably gotten a lot better though.
[1] https://forum.cursor.com/t/how-does-docs-crawling-work/264/3