Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Google / Alphabet will be a strong contender going forward in generative AI. Vertex AI is a pretty decent platform, but Gemini 2.5 pro and Gemini Flash are excellent.

I've been leveraging 2.0 Flash for processing help desk type tickets with long comment chains and its speed / performance is excellent. The only thing I find strange is multi-turn prompts tend to fall apart.



It's hard to bet against Google now, and Sundar deserves his flowers when a year ago everyone was proclaiming the death of Google.

They've caught up and even surpassed Open AI.


When you say "surpassed" do you specifically mean their models are more capable than OpenAI? I'm pretty sure that's what you mean. If that is what you mean, how do you look at it from a product standpoint?

Sounds like ChatGPT has 10 million (paying?) subscribers across their products. Does Google have an analog subscription service ("AI Premium" sounds like one, but I've heard no publicity about their subs)? If they don't have a direct analog, has Google also surpassed ChatGPT with some kind of productization of their AI offerings? If they have, can someone educate me on what that/those products are? I realize AI is, for better or worse, baked into every Google product.

Maybe there're add-ons in, for instance, GCP for those offerings that companies are leveraging, but just trying to understand if this is another case where Google has actually pulled back ahead tech-wise, but still hasn't figured out how to productize it.

edit: grammar


> how do you look at it from a product standpoint?

From what I can tell, (I don't use ChatGPT) Google has every feature OpenAI has, and then a lot more included in other Google products.

Google may not have as many distinct paying AI customers because they get it for free (e.g. college students) or as part of another subscription.


I think Open AI is still far ahead as a consumer choice, and I'm not sure that Google can catch them.

However model capabilities, and especially now that they're integrating Gemini into Google Search (which was a disaster initially, but improving now) the Googlefying is happening.

It'll be curious what they do with projects like Imagen and Veo though.


You can ask Gemini to create images and it calls out to ImageGen, though this sequence may depend on the model you use

Every Google Workspace customer is getting access to the Gemini lineup. That's a pretty big user base


Its easy to bet, more and more people switching to chatbots with tasks they previously used search for, and this can dramatically affect Google main revenue stream: Search Ads.


What’s a multi-turn prompt?


I assume they mean that you send an Array<Message> to the model instead of a single system and user message.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: