Google Is Winning on Every AI Front
Alberto Romero (via Hacker News):
I’d been holding off on writing about Gemini 2.5. Focusing on the AI model didn’t feel like enough to tell the full story of Google’s comeback. Gemini 2.5 is only a piece—albeit a big one—of something much larger. Back in December 2024, I said they would come out on top by the end of 2025. We’re not even halfway there and it’s already happened.
[…]
Perhaps most importantly, the benchmark scores match the signal I receive from vibes checks, high-taste testers, and firsthand testimonials: people are reporting en masse that Gemini 2.5 Pro is indeed the best model today. A rare sight to witness. (Watch Matthew Berman’s clip below.)
And that's just pure performance. Add to the above that Gemini 2.5, compared to models of its category, is fast and cheap—I mean, they're giving away free access!—has a gigantic context window of 1 million tokens (only recently surpassed by Meta’s Llama 4) and it’s connected to the entire Google suite of products (more on that soon).
Previously:
4 Comments RSS · Twitter · Mastodon
Aside from the cringe phrasing of "vibes checks and high-taste testers," I have to agree. My own "vibe check" says they have accelerated very quickly. And they do have the advantage that they basically already have everyone's data, and everyone has an account. An advantage that Apple also potentially had and will continue to squander for the foreseeable future.
People gave them a hard time for a while about "glue on pizza" (sounds kind of old already) but in that span of time they went from that to something that has a substantial lead and is inarguably useful.
Huh. I tried Gemini Pro from December until about 3 weeks ago and it consistently gave me incorrect answers (to engineering class problems) and/or it presented the answers in a very confusing manner. ChatGPT and Grok were almost always correct. Grok wins for outputting the answers in the most understandable format.
Often, I knew the answer that Gemini gave was incorrect right away, so I'd say something like "Are you sure this is correct?" and it'd be like "You're right! My apologies for the oversight. Let me do it again." and it would either output the same wrong answer, or go off on a different wrong tangent.
Or sometimes it would have the equation correct, with the correct numbers, and then it would output the wrong answer. Like 2 + 3 = 5.3 type of stuff.
Guess I'll have to try the new model.
Same. I tried it after reading this post and it never outperformed ChatGPT. Neither in speed nor in answer quality.
"...Or sometimes it would have the equation correct, with the correct numbers, and then it would output the wrong answer. Like 2 + 3 = 5.3 type of stuff..."
BEN G, I have asked for several months now all AI things, does 2 + 2 = 4, and then under what circumstance could it not equal 4. None have replied with anything but 4.
