jeffybefffy519

I wonder if this is a plateau towards the real pricing of AI, if you layer in gemini-2.0-flash at $0.10 / $0.70 then its a 15x price increase to 3.5/3.6 flash. But it hasnt gone up again which is interesting.

The thing i never got about the singularity is that its just one concept that was suggested by a few notable people over time. There is so many different ways AI can go, some which we wouldn't have foreseen.

Whats the bottom here tho? Its obvious what winning is.

Price shock in 3...2...1...

And our solar market regulator is corrupt AF, its just a giant money spinner for battery and solar makers/installer.

Its even different than that, some Codex models like 5.3-codex are terrible at front end work but excel at backend/system design.

I bet it is just GLM 5.2

The actual link https://www.frontiersin.org/journals/neuroscience/articles/1... is fascinating read. Really shows how we have been overly focused on the wrong treatment and not explored some alternative low risk therapies for many conditions.

I dont think gemini needs to be opus level though... if its there to power google search then the scale for google to operate at is enormous compared to openai/anthropic.

Or could you parallelise your experts on different hardware?

IMO Codex has been the same rollercoaster ride as Claude. GPT 5.3-codex was incredible for backend/system tasks, GPT5.5 is better all rounder but weaker in some spots. There has also been many weeks when Codex's models were dumb AF. Same rollercoaster ride as anthropic between Opus 4.5 to 4.8...

IMO the two biggest problems not really being answered by both OpenAI and Anthropic are: 1. Why not make specific models good at specific tasks for Codex/Claude Code. Theres a handful of types of work here whereby small good quality models would do better than these generalised all purpose models whereby someone discovers Fable is bad at biology.... 2. Why cant they consistently run these models and keep them performing? Performance of the models seems to directly correlate with amount of compute available, but they dont talk about it...

I found changing chatgpt's persona settings helped a lot with this!

I havent used fable, but does it cost you the same when it downgrades to opus?

Marketing... that is all

This would probably be the biggest awareness thing tech could do for climate change as well.

I still dont trust the Anthopic and OpenAI are not training on my code. I even just thinking keeping track of what code you have received in prompts and to train/not train on it seems like an impossibly difficult task.

Gemini has done the same thing, gemini-3.5-flash is 15x more expensive for input tokens than gemini-2.0-flash. They are forcing us up the pricing ladder by deprecating the old models....

I wonder if we will see common chiplets that are weights for layers of a model, not the full model just a few common layers that are known for certain things.

Because vibe coding is a toy… thats the secret.

You can use it to accelerate development certainly, but that requires careful change->review cycles. The developer still needs to be in heavy control, versus vibe coding having an agent own the code base.

Its not the right long term solution tho, tiny roof tiles as solar panels have so many problems:

- Magnitude higher number of interconnections which impacts reliability and efficiency

- Uniform roof tile style

- Requires entire roof rebuild which is always more expensive than retrofit of panels on top

- Complex installation resulting in less installers available overall for the market

- Crossing of trades between roofing & electrical

A slightly better solution would have been to make the big traditional solar panels your actual roof panels but really retrofitting them on top of panels solves most of those issues above.

Interestingly if you look at the cost of Gemma3 (this is 12 months old, but demonstrates the authors point) on Vertex AI versus Gemini 2.5 pro, the cost per million tokens WAS very similar.

Just imagine what they do with your data even tho they wont "train on it".... This feels like a "100% beef" moment in history.

Is there a guide you can link for opencode usage like this? I just use codex and find its generally really good. What you are describing sounds like a bit of an unlock.

Surely its just the same model, just allowed to do more work...??

Codex is just as bad with this, i've received two ToS warnings for security research activities so far. I have also tried to appeal with zero response.

Except this isnt using heavily quantised versions of the model thus reducing quality.

Cortical labs have done this before, its their whole thing…

Yup exactly, if this is the truth then put it on the terms/privacy policy etc... exec's say anything these days with zero consequences for lieing in a public forum.

Someone needs to make an actual good benchmark for LLM's that matches real world expectations, theres more to benchmarks than accuracy against a dataset.

I really think the key to addressing climate change should have started 20 years ago in a lot of primary schools, whereby the curriculum includes subjects that are more tailored to solving problems with capturing or converting C02 (eg sciences) so that these students are thinking about these problems when graduating and starting businesses to solve the challenges (hopefully with gov incentives at same time).