Hacker News Reader

Probably a loaded question but how has your 70B model stood against the test of time? I would expect models like Qwen2-72B to outperform yours noticeably? (I get why you can’t simply fine tune that due to the restrictions on that license)

View on HN

Our new 70B is actually completely different from the one we launched a year ago -- we should probably update the naming to make that more clear!

What do you think about its performance on your queries?

Oh interesting, I have to give it a go. I used to use Phind pretty often when it first came out and I told a ton of people about it! I remember the good old days where you offered pretty much unlimited access to ChatGPT/GPT-3. But with your (understandable) quota restrictions and there ended up being so many different services, I haven’t used your service in a while. I had so many subscriptions and the $20 didn’t provide the value I needed at the time. With all these new features, I might revisit it again after my complimentary Perplexity Pro (from my ISP) expires in like 6 months.

(I added that sentence to your text at the top)

Reply on news.ycombinator.com