Comment by nickfromseattle
Comment by nickfromseattle 2 days ago
Side question, let's say Grok is comparable in intelligence to other leading models. Will any serious business switch their default AI capabilities to Grok?
Comment by nickfromseattle 2 days ago
Side question, let's say Grok is comparable in intelligence to other leading models. Will any serious business switch their default AI capabilities to Grok?
I wouldn't for two reasons: - we've already tested OpenAI's GPT and Gemini a lot. Although they're not deterministic, we have used them enough to trust them and know the pitfalls. - Elon's example of Grok outputting 'X is the only place for real information' makes the model almost completely useless for text generation. Even more so than DeepSeek.
Grok 2's per-token prices are similar to GPT-4o, but since Grok tends to write longer responses than others, it can be significantly more expensive to use depending on the task. If xAI prices Grok 3 to compete with o1, not everyone is going to be lining up to use it even if it's a bit better than the competition. If that's how it goes, I'll be interested in the pricing for Grok 3 mini.
> Side question, let's say Grok is comparable in intelligence to other leading models. Will any serious business switch their default AI capabilities to Grok?
Yes, I'd say so. Bear in mind that, outside of the Terminally Online, very few people would deliberately hobble their business by deliberately choosing an inferior product.
The API is compatible, and even if it weren't, it wouldn't matter anyway; everybody has been writing OpenAI API-compatible proxies, including Google, for months now. The only thing that matters is availability, throughput, cost per token (Google is ahead of everyone here: Vertex API is insanely cheap for what it does, Batch API at 50% discount, Prompt Caching at 75%, fully multimodal, better performance in multilingual tasks so actually useful outside the U.S. etc etc etc)
Because people want to protect their businesses?
I don't even think it makes sense to use the closed, API-access models for core business functions. Seems like an existential risk to be handing over that kind of information to a third party in the age of AI.
But handing it over to the guy who is simultaneously running for-profit companies that contract with the government and seizing control of the Treasury? To a guy who has a long history of SEC issues, and who was caught cheating at a video game just to convince everyone he's a genius? That just seems incredibly short-sighted.
? there's tons of businesses and startups building on closed models already
the risk is mitigated by competition in the space and commoditization of LLMs over time. This isn't like building an app solely on Facebook platform or something. It's very easy to swap models. There's platforms that let you use multiple models with the same interface, etc.
We might - I was testing it out on some salesforce apex code and it was doing a better job then o3 mini high at getting to job done for me…