https://security.apple.com/blog/expanding-pcc/ Private Cloud Compute will use Google and Nvidia as outside companies. Apple formed a partnership instead of building its own AI Data Centers; it rents time on Google and Nvidia's infrastructure.
They just conquered mobile hot off losing an antitrust case but getting away with it because AI exists, so they’ve never been in a stronger position. Both Google and Apple are fighting the EU to keep rival AI assistants at bay, they are on the brink of locking everyone else out almost globally.
I think they're concentrating on users and not benchmarks. Gemini is still the best model at plain old searches, like, for parsing images or "what's that poem with this and this in it". It seems to have much more breadth of reasoning than any of the other models, e.g. Gemini 3.1 is still the top model on simple-bench other than Fable.
New 3.5 Flash model is clearly quite optimized for speed. People on this site just look at benchmarks, and underestimate how vital speed is to user experience. No other free model comes close in terms of speed and intelligence, so it seems like the correct strategy to slowly win the mass market, as almost everyone will form their experience of AI through its free version.
I think giving away Gemma may be telling on the direction they wanna go long term.
Winning doesn't have to mean making the most money I hope. Maybe winning is paying homage to the years of organizing everyone's data and giving it back in the form of a highly optimized completely free capable model? Or am I too naive. Give a highly optimized code forge I can run on local affordable hardware and you will win in my eyes.
Giving away Gemma was the first time I thought google may have a heart after all. Or am I too naive.
That was probably their low, and they've been catching up especially as the field has plateaued. They're making contracts to host LLM services, so they don't seem to be giving up, and the longer the plateau lasts, the less they'll be behind, unless they are slow to migrate to whatever the next step is.
I don't think Google has given up; rather, I believe they are taking a wait-and-see approach and carefully controlling their investment. Furthermore, I feel Gemini is still excellent at simple, short-term, and even multimodal tasks, but it is not well-suited for long-term planning scenarios for AI agents.
10 comments
[ 0.20 ms ] story [ 14.1 ms ] threadNew 3.5 Flash model is clearly quite optimized for speed. People on this site just look at benchmarks, and underestimate how vital speed is to user experience. No other free model comes close in terms of speed and intelligence, so it seems like the correct strategy to slowly win the mass market, as almost everyone will form their experience of AI through its free version.
Winning doesn't have to mean making the most money I hope. Maybe winning is paying homage to the years of organizing everyone's data and giving it back in the form of a highly optimized completely free capable model? Or am I too naive. Give a highly optimized code forge I can run on local affordable hardware and you will win in my eyes.
Giving away Gemma was the first time I thought google may have a heart after all. Or am I too naive.
Yes
That was probably their low, and they've been catching up especially as the field has plateaued. They're making contracts to host LLM services, so they don't seem to be giving up, and the longer the plateau lasts, the less they'll be behind, unless they are slow to migrate to whatever the next step is.