oezi

8 ms·

91.
by oezi 6mo ago
I recently looked at this a bit but came away with the impression that at least on API pricing the models should be very profitable considering primarily the electricity cost. Subscriptions and free plans are the thing that can easily burn
92.
by oezi 6mo ago
This! And with AI studio you get a couple of free calls per day (it has gotten less and less). I have had days where I would be able to get 100 USD worth of tokens from AI studio for free. 1m tokens in and great code out.
93.
by oezi 6mo ago
The article is certainly firebranding, but the core tenet strikes a valid point: how has the US lost the plot within such a short time? How did it go from the flag bearer of freedom and progress to isolationist bully that wants to invade Gr
94.
by oezi 6mo ago
I think I would love to test it, but on the Pro plan I just did two small sessions with 4.6 Sonnet and it consumed my 5h quota within one hour.
95.
by oezi 6mo ago
The tokenizer changes seem to indicate that 4.7 isn't just a checkpoint but rather a model trained mostly from scratch, right?
96.
by oezi 6mo ago
> — billing alerts fire in hours, damage happens in minutes. And why do you need to use AI to tell us that. How much shorter could the prompt have been?
97.
by oezi 6mo ago
I think after the TTL expires the session should be autocompacted and the user should given a choice to continue with compacted version or be hit with the full read cost of continuing with their large but expired context. At the moment user
98.
by oezi 6mo ago
On OpenRouter token consumption is up 5x since November 2025. If this is indicative of the industries growth then I can't fathom how we will not hit resource constraints.
99.
by oezi 6mo ago
I tried it, but it doesn't make for good gameplay, it just gets too easy. Could maybe subtract points, but that also feels strange. I updated gravity to do things, but orbiting isn't permitted.
100.
by oezi 6mo ago
True. Hard to square it being a game, fast-spaced and accurate.
101.
by oezi 6mo ago
19:50 Put codex and claude (thinking high) to work in parallel to see who could come up with the better physically accurate mindless tapping orbital mechanics sandbox. 20:10 Both codex and claude finish pretty much at the same time, but my
102.
by oezi 6mo ago
If in Sydney: https://maps.app.goo.gl/q9symY7GrNh8GRNe8
103.
by oezi 6mo ago
Instead of using just positional encodings, we absolutely should have speaker encodings added on top of tokens.
104.
by oezi 6mo ago
Coding assistants won't win this game. They sure will win the hearts of developers, but to scale you need mass adoption and products for which users want to pay substantially. OpenAI is falling behind in the small features in their cha
105.
by oezi 6mo ago
Looks neat, thx. It is really interesting how many facets the same kind of issue can produce. I think using expect from baker is a great way to deal with some steps that are hard to automate otherwise.
106.
by oezi 6mo ago
I see absolutely no upside for a society to allow sports betting. The tax revenues don't justify the addiction, debts and devastated families.
107.
by oezi 7mo ago
They don't mention a twinkle that many task runners seem keen to omit: how do you handle things where there are human steps involved and not everything is automated? How do you track what has worked and what is still left to do if thin
108.
by oezi 7mo ago
Just a warning that plain WhisperX is more accurate and Whisper-timestamped has many weird quirks.
109.
by oezi 7mo ago
I think Stackoverflow.com should have pivoted to something similar. Let AIs both pose, answer and vote on questions and answers.
110.
by oezi 7mo ago
Tell me where you are from without telling me where you are from... Poland I guess?
111.
by oezi 7mo ago
Well in 2 years AI will have so advanced that none of this matters.
112.
by oezi 7mo ago
Is there a Autoresearch for Jupyter somewhere? I point it to a Jupyter cell to improve based on another which calculates the target metric?
113.
by oezi 8mo ago
Do you also support timestamps the detected word or even down to characters?
114.
by oezi 8mo ago
I think the primary issue is that the metric is tokens and not dollars spent.
115.
by oezi 8mo ago
> Schuldstolz Never heard that word before. And I don't think I am lecturing you about something you should do. I was just talking about why Germans in their own free country are choosing to make decisions about their own laws. If y
116.
by oezi 8mo ago
It is often easy to assume this position if you are majority, white, employed, etc. Your argument is similar to saying that we shouldn't have rules when driving cars. "Why life cannot be dignified if I have to observe stop signs.&
117.
by oezi 8mo ago
A little less hyperbole would maybe help your arguments, but trying to argue that one of the most liberal democracies in the world is comparable to one of the most repressive regimes is hurting your argument ( https://ourworldinda
118.
by oezi 8mo ago
That isn't really what we perceive (at least if educated). We see that Free Speech is not an absolute right, but is secondary to the most important right which for Germans is Human Dignity. It might be foreign to you because your const
119.
by oezi 8mo ago
Agreed
120.
by oezi 8mo ago
The key thing to understand is that Europeans want clear rules around hate speech, online harassment and such. Thus lawmakers are acting to find laws which encapsulate these. In Germany, we have some simple ones surrounding using Nazi symbo

More ›