@CliffordAsness I'm a centrist and think poor people shouldn't be allowed to get credit, besides narrowly for cheap cars and car repairs. And since AVs are going to make taxis cheaper than ownership, even that narrow point is going away.
@orrdavid@CliffordAsness The difficulty is differentiating the transiently poor from the congenitally poor. If it were legal for lenders to underwrite based on IQ they would do it, and all issues would be solved.
@wizplum@PauloMacro Isn't the point of a greenshoe that the issuer trades a bit of dilution for a bit of price stabilization, & if the price stabilization increases demand, it's worth it? The stabilization can be via issuance into a pop or open market purchases in a fade, either way it stabilizes.
Agreed. I loved the way GPT speaks, but Neon is an even better conversational partner. More felicitous phrasing and seemingly more willingness to push back, which in addition to being desirable in itself makes it feel more engaging.
Gemini 4 Neon vs Opus 5.5 Review
I am going to start this by saying I still think Opus 5.5 is the best ALL-AROUND model you can use right now and the usage limits are still better in Claude code.
Now that I’ve said that Neon is actually the model I prefer to use for my day-to-day work.
Communication:
Neon wins this by miles. It is so nice to talk to this model. It communicates complex issues in a way no other model does. This makes understanding what it’s doing or changes you should make so much easier.
Coding:
I’ve been pushing Neon super hard here and man it just keeps impressing me. I think it is hands down the best coding model out there. Not only is it good but it helps you fully understand what it did which is even better.
Front End/Taste:
Opus still wins here as Claude models just have that extra something that makes them amazing at front end design. Admittedly I need to test more with Neon here but the front end work I’ve done is good but does not beat Opus
Knowledge work:
Google’s training data has clearly been baked into Neon and it’s depth of knowledge about any subject is clear.
3D Modeling:
Opus wins here. Neon is not great at 3D demos which kind of sucks as I really like the models I use to be able to do this but it’s by no means a super important capability.
Trust:
This is the unmeasurable and it’s the main reason I still use Opus a lot. I have learned to trust Opus because I’ve used it for so long so I know what it’s capable of and where it will need extra instruction. I don’t fully trust Neon yet. I haven’t spent enough time pushing the limits of what the model is capable of to say I trust it fully as my day to day work horse. The one thing I can say is everything I throw at neon it absolutely crushes with speed and then it comes back with a concise summary of what it did and why.
As time goes on I will begin to transfer more and more of my work to Neon. I truly feel like Neon is a partner that understands intent and talks to me in the way I’d want a teacher to. I congratulate the entire deepmind team on creating this model it truly is a joy to use.
@wizplum But I wonder if they also saved by just being a fast follower: let OAI/ANT torch capital learning what works, then implement those learnings once progress seemed saturated.
Ok, Gemini 4 Neon is very impressive, at least for my unsophisticated use. And since I already pay for a Google workspace, I get it free. Seems bad for $ANT
The compute credits in the Gemini Ultra sub are insane. The $100 plan gets you 1000 colab credits/month. That's ~55 H100 or 100 RTX6000 Pro (G4) hours.
The colab CLI let your agent to use these credits like Modal... no notebooks required!
Come for the Neon, stay for the GPUs.
Oracle has secured all the funding we need to fulfill our contract with OpenAI. With our final capital raise earlier this month, we became the first hyperscaler fully funded for this investment cycle. The financing we outlined at Oracle Investor Day, including vendor financing, is in place. Any further private discussions with vendors are just additional options. Our ability to deliver on our OpenAI contract does not depend on those discussions. Any media reports that will come out and say the contrary are incorrect and wrong.
Oracle and OpenAI are making rapid progress on our infrastructure partnership. Our Abilene campus, where Astra-6 was trained, is nearly complete. Shackelford, our next gigawatt-scale campus, is fully permitted. The handover of NVIDIA Vera Rubin racks in its data halls is on track for the coming weeks.
We continue to expect positive free cash flow in the second half of FY2028. We are targeting more than $30 billion in free cash flow in FY2029 and expect to self-fund future growth beginning that year.
@glbeaty@atelicinvest@SowingAlphaSeed Is there reason to believe that distillation has been thwarted rather than merely slowed? If it takes six months rather than six weeks to distill a model, that still seems like a disaster for frontier labs.