91–100 of 332 posts
I rarely used Codex compared to Claude because it was extremely slow in GitHub copilot . Like maybe 2-5X slower than Claude Sonnet. I really wish they just made their models faster than “better”
so this was arctic fox it seems, lot of us ended up downgrading to codex 5.0 because of the token burn was too much, i see codex max is a step up which is welcome but still unsure if they solved that github issue around tool use that impacts tokens going to wait and see after being burned by 5.1 before i upgrade back to 0.58 gemini 3 has been a let down tbh to see agentic coding wasn't a top priority im sticking with…
I've been using a lot of Claude and Codex recently. One huge difference I notice between Codex and Claude code is that, while Claude basically disregards your instructions (CLAUDE.md) entirely, Codex is extremely, painfully, doggedly persistent in following every last character of them - to the point that i've seen it work for 30 minutes to convolute some solution that was only convoluted because of some sentence I t…
> Claude basically disregards your instructions (CLAUDE.md) entirely A friend of mine tells Claude to always address him as “Mr Tinkleberry”, he says he can tell when Claude is not paying attention to the instructions on CLAUDE.md when Claude stops calling him “Mr Tinkleberry” consistently
I've vibe coded Godot games extensively.
Just about every model I've tried likes to invent imaginary functions.
I was really prefer for there to be a way for me to pick model trained in whatever framework I need.
Reviewing AI generated code feels like editing a long book, and every now and then you notice some words are just completely made up. You then ask the AI to fix its book, and it will just add more AI generated words.
On one hand I want this to be a reality check to everyone who's trying to lay off real software engineers to replace us with AI.
On the other hand half of the stock market is held up by overhyped AI valuations. If the tide goes out too fast, and there is a mass realization that this stuff just isn't as good as it's hyped to be, it's not going to be fun for anyone.
I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…
Please give me an option for a password (or passkey) or literally anything else that doesn't require either linking with google or going through an email flow for every login
> Compaction enables GPT‑5.1-Codex-Max to complete tasks that would have previously failed due to context-window limits, such as complex refactors and long-running agent loops by pruning its history while preserving the most important context over long horizons. In Codex applications, GPT‑5.1-Codex-Max automatically compacts its session when it approaches its context window limit, giving it a fresh context window. It…
Attention is quadratic, so you have to pick a cutoff for context window size. In addition, the error/noise in state space increases with longer contexts, resulting in poorer performance. So even if you're willing to take the O(n^2) slowdown of a larger context window, it still won't work.
Exactly. Standard Multi-Head Attention uses a matrix that grows to 4B parameters for a 64K sequence as a starting place. FlashAttention v2 helps slightly, but as you grow to 128K context length, you still need over 1TB/s memory bandwidth to stay compute-bound in practice even with this optimization.
So there has been a lot of research in this area and model architectures released this year are showing some promising improvements. Sliding windows lose context fidelity and if you go fully linear, you sacrifice math, logic, and long multi-turn (agentic) capabilities, so everyone is searching for a good alternative compromise.
MiniMax-M1 had lightning attention to scale up to 1M context lengths. It's "I/O aware" via tiling and calculates attention two ways block-wise (intra-block traditional attention and inter-block linear attention), thereby avoiding the speed-inhibiting cumulative summation.
DeepSeek V3.2 uses DeepSeek Sparse Attention (DSA), which is sub-linear by only computing "interesting" pairs. For example, in 128K context lengths this requires only 10-20% of attention pairs to be materialized.
Both Qwen3-Next and Kimi Linear adopt a Gated DeltaNet, which is borrowed from Mamba2. In Qwen3-Next it alternates three Gated DeltaNet (linear attention) layers for every one gated [full] attention. The speedup is from a delta rule, which basically amounts to caching in a hand-wavy way.
There's no universally-adopted solution yet, as these are all pretty heavy-duty compromises, but the search is going strong right now for linear or better attention mechanisms that still perform well.
I really would prefer them to start creating customized models. I've vibe coded Godot games extensively. Just about every model I've tried likes to invent imaginary functions. I was really prefer for there to be a way for me to pick model trained in whatever framework I need. Reviewing AI generated code feels like editing a long book, and every now and then you notice some words are just completely made up. You then…
That was annoying back then, but these days that's not so much of a problem.
You can write your program and then simply have it invent the library as well, while it's at it! ;)
Thinking level xhigh: https://tools.simonwillison.net/svg-render#%20%20%3Csvg%20xm...
I would love to see all the big players put 1% of the effort they put into model training into making the basic process of paying and signing in suck less. Claude: they barely have a signin system at all. Multiple account support doesn’t exist. The minimum seat count for business is nonsense. The data retention policies are weak. OpenAI: Make ZDR a thing you can use or buy without talking to sales, already. And for t…