Hacker News Reader

Comment by CuriouslyC | Hacker News Reader

This is pretty sweet. The speed is nice but what I really care about is you bringing the per token cost down compared with models on the level of mistral medium/gpt4. GPT3.5 is pretty close in terms of cost/token but the quality isn't there and GPT4 is overpriced. Having GPT4 quality at sub-gpt3.5 prices will enable a lot of things though.

View on HN

Mixtral's quality is definitely up there with Gpt3.5. Specifically for coding, i consider them almost equivalent in quality. In fact Mixtral 8x7 is starting to be my go-to coding assistant instead of Gpt. It is fast, it is accurate, and i think i like his responses better than Gpt.

Reducing LLM size almost 10 times in the span of a little more than a year, that's great stuff. Next step i think is 3 billion parameters MoE with 20 experts.

What's the difference in your own words/opinion in quality between GPT-3.5 and GPT-4? For what usecases?

GPT3.5 is great at spitting out marketing babble, summarizing documents and performing superficial analysis but it doesn't take style prompts as well as gpt-4 and its reasoning is significantly worse when you want it to chain of thought follow a complex process while referencing context guidance.

I wonder if Gemini Pro 1.5 will act as a forcing function to lower GPT4 pricing.

Is that available via an API now?

Kind of, it's in a "Private Preview" with a waitlist.

And in non EU countries.

Via GCP only?

GPT-4 is overpriced vs what?

You seem to be implying that Mistral Medium is on the same level as GPT-4?

Reply on news.ycombinator.com