playing with Simon Willison’s LLM CLI tool - https://llm.datasette.io/en/stable/
seems quite powerful!
Posts
@derekyt_ @LivDaBanana @ulintelligence oh interesting lol
how are people replying to him then? i literally don’t see the reply at all
@chrisparkX - grok is so much more expensive than any other “smart” LLM, and the cheaper ones often get better results
- mechahitler
- GPT-5 is much more accessible due to the cost, and the speed
@paytondev_priv aww
@NeoFreeBird I have to wait over a week to install it 😭😭😭
Here’s a good “For You” feed too → https://bsky.app/profile/did:plc:3guzzweuqraryl3rdkimjamk/feed/for-you
They’re offering developers foundation models that already exist on-device for any user with the latest iPhone, and the user won’t need to download any more models for a specific app.
It’d be cool to see other OSs do this imo.
BUT if this happens, your local storage will be super full of compartmentalised LLMs that are in their own directories on a per-app basis.
This is why I think what Apple is doing with Apple intelligence is quite promising.
Been thinking about these open weight models and the idea of running it on a phone/laptop (any of your own devices) and I feel like it’s cool if separate apps that currently offer LLMs in the cloud will start supporting local LLMs (speed, free, private) within their apps… 🧵
@sama @OpenAI 👌