AmazingTurtle

5 ms·

1.
by AmazingTurtle 2mo ago
> Could there be a benefit to releasing a new model, slowly dumbing it down over a couple months, then releasing a new model that’s marginally if at all better than the original to create a perceived improvement when in reality there isn
2.
by AmazingTurtle 2mo ago
This is the first time that I hope that china fucks this up
3.
by AmazingTurtle 2mo ago
You better bet someone started their agents with a prompt "Make a salesforce clone but with 100% uptime"
4.
by AmazingTurtle 2mo ago
thought this link is better / cleaner: https://unwall.app/www.economist.com/science-and-technology/...
5.
by AmazingTurtle 2mo ago
Yeah only somtimes it rewrites the whole files and thats the models/developer instructions fault. a good AGENTS.md is all you need, change my mind
6.
by AmazingTurtle 2mo ago
still half as cheap as in EU
7.
by AmazingTurtle 2mo ago
gpt-6-astra is a bitch, it constantly scope creeps itself with "yet another thing" to give it that darn polished lick. the results are eventually a little bit better but at what cost? let's do the math. gpt-5.6-sol: 1x base g
8.
by AmazingTurtle 2mo ago
I consider myself an advanced industry representative in these things and would also borrow some time for detailed investigation. I'd happily contribute some valuable assets (tacos with cheese dip) into the matter, straight out of my p
9.
by AmazingTurtle 2mo ago
I use cachyos as my daily driver for a long time now. I play quite a few games. Only league of legends forces me to dual boot into windows every once in a week or so - fuck the anticheat cartel
10.
by AmazingTurtle 2mo ago
Yeah that must be it. OpenAI doesn't want to disclose internal reasoning, that's why thats typically encrypted_content in OpenAI codex session ledgers etc.; leveraging responses API preserves reasoning server side all the way till
11.
by AmazingTurtle 2mo ago
French
12.
by AmazingTurtle 2mo ago
Well, guess they are just trying to conform all the weird EU checkmarks
13.
by AmazingTurtle 2mo ago
There are already a handful of repos doing essentially exactly this: `mlx-moe-offload`, `streamlx`, `mlx-moe`, `mlx-flash`, and `deepseek-v4-flash-mlx` - i.e. keep the resident parts of an MoE in unified memory and page/stream routed e
14.
by AmazingTurtle 3mo ago
But I'm fine paying +50% more for +5% increased performance because it will pay off.
15.
by AmazingTurtle 3mo ago
that model is 14MB large what do you expect. but I agree it's funny regardless
16.
by AmazingTurtle 3mo ago
↲ is also two tokens instead of a simple \n lmao
17.
by AmazingTurtle 3mo ago
So it's basically a container with a fancy name, innit?
18.
by AmazingTurtle 3mo ago
how did this make it to the front page
19.
by AmazingTurtle 3mo ago
coughs in agressive nord vpn ads
20.
by AmazingTurtle 3mo ago
75% quality... it gets worse the longer the context.
21.
by AmazingTurtle 3mo ago
Really just a (not so) fancy CLI wrapper about a prompt and a skill: https://github.com/openai/codex-security/blob/f22d4a36f26d16...
22.
by AmazingTurtle 4mo ago
I have a suspicion that they are just trained on puzzles by now
23.
by AmazingTurtle 4mo ago
Funny timing. I've been building something similar in my spare time called Daydream. There’s a lot of overlap: local screen/audio capture, OCR and transcription, window and activity context, SQLite, and a searchable memory of what
24.
by AmazingTurtle 4mo ago
codex CLI is already leveraging this internally. instead of performing raw MCP calls its internally using a js REPL to do that. elicitation is then propagated for approvals independently / it's interpreted as regular tool calls fo
25.
by AmazingTurtle 4mo ago
This happened to me one time. On windows though, deleting C:/ lmao. The direction I'm heading now is a better harness, i.e. isolating codex at a container level with dedicated workspaces / mounts etc. I'm building someth
26.
OpenAI reduces Codex Model Context Size from 372k to 272k(github.com)
371 points by AmazingTurtle 4mo ago|166 comments
27.
by AmazingTurtle 4mo ago
Lmao this goes for literally any model. Deepseek etc is Chinese models and OpenAI/Anthropic/xAI are very western.
28.
by AmazingTurtle 4mo ago
I learned that running goal for hours produces exponentially more slop than running targeted prompts over and over again manually. Personally, I use gpt 5.5 high with planning every time and plan various smaller features/changes in par
29.
by AmazingTurtle 4mo ago
It's funny, they sell you a subscription for frontier models, then over time begin to nerf them rapidly and no one talks about it. Should give me a discount when they reduce reasoning effort silently on the server side! But on the othe
30.
by AmazingTurtle 5mo ago
how can one be sure you don't do rugpull in the future?

More ›