HN user
China never threatened this: "A whole civilization will die tonight, never to be brought back again.". Also China never announced that it was going to attack Europe. I trust them much more compared to a malignant narcissist that doesn't care if the whole world burns now that he doesn't have much left to live.
Suddenly burning up the quota ~4x faster than usual is not a mess up in your opinion?
Claude Code cache is not 1 hour. There is a "Closed as not planned" issue in GitHub that confirms that it has been moved to 5 minutes since March: https://github.com/anthropics/claude-code/issues/46829. I started seeing the massive degradation exactly on the 23rd of March, hence after a few days I unsubscribed because it was completely unusable, with a ~5h session being depleted in as little as 15-20 mins.
If you are speaking about the world, hundreds of millions in the next 5 years is probably closer to reality in my opinion. And from your question I think that you already know the answer.
What change did you release on March 23rd when the subscription limits collapsed and they are still way down compared to what they used to be?
How many humans have been born until now and how many Einsteins have been born? And in how many hundreds of thousands of years?
What about NVDA (~$4.5T) and AVGO (~$1.8T)?
How much more was worth USD at the beginning of the year?
Flux kontext quality is noticeably worse that nano banana, Qwen image 2509 and Seedream 4 most of the times. For pure image generation instead Hunyuan image is scarily good.
Seedream 4 is better than nano banana on average, so that test result seems accurate to me
The biggest model that they have used has only 760M parameters, and it outperforms models 1 order of magnitude larger.
Yeah, less than from Milan to my city still in Italy…
I’m not an expert, but I would say extremely small.
For comparison Hunyuan video encodes a shit-ton of videos and rudimentary real world physics understanding, at very high quality in only 13B parameters. LLAMA 3.3 encodes a good chunk of all the knowledge available to humanity in only 70B parameters. And this is only considering open source models, the closed source one may be even more efficient.
Exactly. The previous version of o1 did actually worse in the coding benchmarks, so I would expect it to be worse in real life scenarios. The new version released a few days ago on the other hand is better in the benchmarks, so it would seem strange that someone used it and is saying that it’s worse than Claude.
As I said, o3 demonstrated field medal level research capacity in the frontier math tests. But I’m sure that your use cases are much more difficult than that, obviously.
Where is the plateau? Chatgtp 4 was ~0% in ARC-AGI. 4o was 5%. This model literally solved it with a score higher than the 85% of the average human. And let’s not forget the unbelievable 25% in frontier math, where all the most brilliant mathematicians in the world cannot solve by themselves a lot of the problems. We are speaking about cutting edge math research problems that are out of reach from practically everyone. You will get a rude awakening if you call this unbelievable advancement a “plateau”.
You must be stuck at SDXL for posting something absolutely and verifiably false as the sentence above.
Yeah.. generations. I really hope that it doesn’t end like the New York Times article saying that human flight was at best hundred of years away a few weeks before the Wright brothers flight..
Absolutely false. Trump got 74M votes in last election in 2020. US population was 330 millions. Today US population is ~340M. I seriously doubt that more than 170M people support a convicted felon for president. I bet that he will get less than 170M votes in less than a month. Do you want to take the other side of the bet to back up your provably wrong assertion?
Good call, here cannabis damaging effects are usually minimised and hidden under the rug. And you will always, always find someone telling you “what about alcohol”
All the nuclear energy production plants in the world ever built have the same capacity as the amount of solar generation built just last year. So your comparison doesn’t make much sense, nuclear is dead in current energy market, it makes sense only for producing fissile material for nuclear bombs.
There was Sam Altman taking live prompt requests on twitter and generating videos. They were not the same quality as some of the ones in the website, but they were still incredibly impressive.
Feeling the heat? Did you actually watch the videos? That was a huge leap forward compared to anything existing at the moment. Order of magnitudes away from a blog post discussing a model that maybe will finally be on par with chat gtp 4...
Yes, they were. They got stuck with ice and they disagreed in the air speed sent. Because of this the autopilot was disabled and the flight controls were switched in alternate law 2. The pilot that was flying failed to realise that this meant that his inputs had a much bigger effect compared to normal flight and he panicked and he also failed to relinquish control to the much more experienced pilot (on that model) multiple times. How are the pitot tubes not the root cause?
No way that they are going to ground them now. Last time they were delaying the grounding as much as possible even after the second fatal crash. They grounded it only after a lot of other countries already did it.
April 2023 is for gtp-4 turbo (aka 1106-preview)
Why not? You don’t need always peak power, in the night when there is no solar production the peak power consumption is much lower. And even accounting for average generation, using the capacity factor of 20% for solar and 90% for nuclear it means that you would need 4 years and half at current production level to deploy the entire average energy produced by all the nuclear power plants in the world. But given how the solar deployments are increasing YoY it will be probably closer to 2y 1/2 or 3y.
You are completely wrong. You don’t need only 2 or 3 nuclear power plants.
You would need to build again the total global nuclear capacity ever built only to balance the 413GW of solar deployed just this year.
Nuclear is dead, whoever says otherwise doesn’t know anything about the current energy market or is in denial.
Nuclear can’t simply be built as cheaply and fast as solar.
https://www.iea.org/energy-system/electricity/nuclear-power# “ Nuclear energy, with around 413 gigawatts (GW) of capacity operating in 32 countries”
So the total global nuclear capacity ever built is now producing 413GW. Eerily the same as the solar capacity installed just last year.
Whoever is pushing for nuclear instead of solar in 2023 either doesn’t have the slightest idea about the current energy market or simply wants to help the fossil fuels to survive given how long would it take to build a nuclear power plant and at what cost.
And yet spacex was profitable this year. So if it’s true that they are spending $2B per year on starship while being profitable it’s unbelievably good for them.