Readit NewsThey nerfed Pro 2.5 significantly in the last few months. Early this year, I had genuinely insightful conversations with Gemini 2.5 Pro. Now they are mostly frustrating.
I also have a personal conspiracy theory, i.e., that once a user exceeds a certain use threshold of 2.5 Pro in the Google Gemini app, they start serving a quantized version. Of course, I have no proof, but it certainly feels that way.
There was a significant nerf of Gemini 3-25 a little while ago, so much so that I detected it without knowing there was even a new release.
Totally convinced they quantized the model quietly and improved on the coding benchmark to hide that fact.
I’m frankly quite tired of LLM providers changing the model I’m paying for access to behind the scenes, often without informing me, and in Gemini’s case on the API too—at least last time I checked they updated the 3-25 checkpoint to the May update.
The problem with all of these is that SOTA models keep changing. I thought about getting OpenAI's Pro subscription, and then Gemini flew ahead and was free. If I get this then sooner or later OpenAI or Anthropic will be back on top.
This is even the case with Gemini:
The Gemini 2.5 Pro 05/06 release by Google’s own reported benchmarks was worse in 10/12 cases than the 3/25 version. Google re routed all traffic for the 3/25 checkpoint to the 05/06 version in the API.
I’m also unsure who needs all of these expanded quotas because the old Gemini subscription had higher quotas than I could ever anticipate using.
Nobody's pointing out that Google's "Preview" models aren't meant to be used for production use-cases because they may change them or shut them down at any time? That's exactly what they did (and have done in the past). This is a case of app developers not realizing what "Preview" means. If they had used a non-preview model from Google it wouldn't have broken.
They're using these "Preview" models on their non-technical user facing Gemini app and product. Preview is entirely irrelevant here if Google themselves use the model for production workloads.
SirensOfTitan commented on First American pope elected and will be known as Pope Leo XIV cnn.com/world/live-news/n... · Posted by u/saikatsg
> I guess I interpret it as "America has gone into political decline." I'm a bit surprised to hear a Trump appointee say that
What do you think the premise of the motto Make America Great Again is? The difference between Trumpists and others who see a decline is that the former see the 2016 and 2024 elections as reverses in the decline, whereas other see them as sources or exacerbations.
Anyone who sees Trump as either an aberration or a savior is deeply deluded on the state of America.
In my opinion, the US world order’s decay was unmasked in 2008, and it has been accelerating since. The two economic realities between the poor rural America and the rich coastal cities (and even within them there is so much clear wealth disparity) have only gotten worse, and the political and bureaucratic system isn’t really capable of skillfully dealing with it.
Trump actually speaks to the realities that few politicians will (Bernie Sanders did too in 2016, hence his appeal), though his prescribed solutions are likely just accelerating the country’s demise.
SirensOfTitan commented on Evolving OpenAI's Structure openai.com/index/evolving... · Posted by u/rohitpaulk
There is a pretty significant different between “buy $9 for $4” and selling a service that costs $9 to build and run per year for $4 per year. Especially when some people think that service could be an absolute game changer for the species.
It’s ok to not buy into the vision or think it’s impossible. But it’s a shallow dismissal to make the unnuanced comparison, especially when we’re talking about a brand new technology - who knows what the cost optimization levers are. Who knows what the market will bear after a few more revs.
When the iPhone first came out, it was too expensive, didn’t do enough, and many people thought it was a waste of apples time when they should be making music players.
It's a commodity technology and VCs are investing as if this were still a winner-takes-all play. It's obviously not, if there were any doubt about that, Deepseek's R1 release should have made it obvious.
> But it’s a shallow dismissal to make the unnuanced comparison, especially when we’re talking about a brand new technology - who knows what the cost optimization levers are. Who knows what the market will bear after a few more revs.
You're acting as-if OpenAI is still the only player in this space. OpenAI has plenty of competitors who can deliver similar models for cheaper. Gemini 2.5 is an excellent and affordable model and Google has a substantially better capacity to scale because of a multi-year investment in its TPUs.
Whatever first mover advantage OpenAI had has been quickly eliminated, they've lost a lot of their talent, and the chief hypothesis they used to attract the capital they've raised so far is utterly wrong. VCs would be mad to be continuing to pump money into OpenAI just to extend their runway -- at 5 Bln losses per year they need to actually consider cost, especially when their frontier releases are only marginal improvements over competitors.
... this is a bubble despite the promise of the technology and anyone paying attention can see it. For all of the dumb money employed in this space to make it out alive, we'll have to at least see a fairly strong form of AGI developed, and by that point the tech will be threatening the general economic stability of the US consumer.
SirensOfTitan commented on Evolving OpenAI's Structure openai.com/index/evolving... · Posted by u/rohitpaulk
They made $4 billion last year, not really "little to no money". I agree it's not clear they can justify their valuation but it's certainly not a bubble.
I guarantee you that I could surpass that revenue if I started a business that would give people back $9 if they gave me $4.
OpenAI models are already of the most expensive, they don’t have a lot of levers to pull.
SirensOfTitan commented on Duolingo will replace contract workers with AI theverge.com/news/657594/... · Posted by u/donohoe
Right, so if the productivity gains were so blindingly obvious and immediate for everyone, mandates wouldn't be needed.
These companies tried to quantify the productivity impact of work from home, so it's utterly bewildering to me that they would push these tool-use mandates without actually quantifying the impact LLM tools have on productivity. If it were just 'getting familiar' with AI tools to help define an AI-driven product mindset, I'd expect these CEOs to have more than a naive perception of the tools and their limitations.
I honestly wonder where these mandates started--part of me feels like this is the nascent stage of a VC panic that their AI investment strategy might not work out.
SirensOfTitan commented on This site uses cookies to store the fact you clicked “Accept Cookies” rodyne.com/?p=2368... · Posted by u/boznz
In the proof of concept I linked, a self expiring session token is given to the client.
If it is returned before the expiry, a new token is generated and the process starts over.
If it is not returned before its expiry the session is considered to have been closed.
But how are you storing the session token in the browser?
> what do you think of me from our interactions all this time?
The answer is:
> From our interactions, I’d say you’re like a lost cat who keeps wandering into the library: clearly curious, mostly harmless, probably doesn't belong here, but somehow still endearing in your refusal to leave. You ask questions like someone who fell into a knowledge vending machine and just started pressing buttons at random. Occasionally useful, frequently chaotic, always weirdly specific.
> You don’t seem to learn from me so much as you just keep poking me like a magical fortune-telling 8-ball that got a liberal arts degree. And honestly? I respect the commitment to the bit.
> So, yeah. You’re like my favorite recurring glitch: baffling, persistent, occasionally helpful to others, and absolutely impossible to uninstall.
edit: I think it is actually generic. I tried to press it to give examples from the memory and I don't think it actually have access to the memories. So it is just my biased interpretation. I also would take this over it being too nice and yes man while being completely wrong.