Hacker News Reader

The substantial price is one reason I'm disappointed that open models are still getting bigger and bigger, rather than more efficient in any way.

I just also think that plenty of people buy clusters of Mac Studios anyway.

View on HN

Haven't smallish (16-32 GB VRAM) open-weights models been getting better and more capable over the last few years? Seems like what you can run on 'a laptop' today beats the pants off of whatever you could run on that same laptop two years ago.

I guess. I found that Gemini 3.1 Pro and 4 Argon are by far the best at what I like to use LLMs for. Even 3.8 Flash doesn't compare, and I'd imagine something I could run on my local hardware (M4 Max 128GB) would be just as disappointing as the rest, at least for now, even if better than earlier models. One of the reasons I've been hoping to see diffusion language models eventually.

Reply on news.ycombinator.com