@MatMilbury @thdxr You never die by pressing red. Though you most definitely would deserve to die given your reading comprehension skills
English
Nemo
78 posts
@thdxr Why crazy?
Pressing blue: I'm going to die or live in a society that believes we can do great things by coordinating effort
Pressing red: I die or live in a society of selfish pricks
English
@slimjimmy LLM can solve open math problems, that single digit IQs like you can’t even understand, though
English
where i'm at on LLMs:
1. LLMs are NOT going to outmode software engineers, not now, not ever
2. LLMs alone are a dead end for attaining anything like AGI
3. forget about LLMs achieving ASI
4. LLMs will only make good engineers faster, marginally
5. LLMs will remain poor at architecture and design
this is 100% because LLMs cannot reason
and no, generating a bunch of hidden context is not "reasoning"
as always, when the same people are making claims like this, have a look at what they stand to gain
and to them, i warn directly: you are going to fall. hard
English
@giangnguyen2412 @JulieKallini Why don’t you follow your brilliant strategy and ask them yourself, Sherlock? I’m going to enjoy the show
GIF
English
@JulieKallini Why just not ask them directly? Why all these conjectures?
English
1/ "New tokenizer" does not imply "new base model," and "new base model" is not the simplest explanation. There are much simpler explanations that fit Anthropic's public description of Opus 4.7 equally well.
Nathan Lambert@natolambert
Opus 4.7 has a new tokenizer. This means it's also a new base model. Glory days of pretraining still very much going.
English
@RihardJarc Plot by itself doesn't mean much, unless you also show the TAM growth in the same period
English
OpenAI continues to lose market share in GenAI website traffic, while $GOOGL Gemini, and Claude are gaining:
- ChatGPT 56.72% vs 77.43% 12 months ago
- Gemini 25.46% vs 6% 12 months ago
- Claude 6.02% vs 1.4% 12 months ago
At this point in the race its all about distribution & the cost of serving these models.
English
@b_shulha @flaviocopes Reading comprehension issues. It's actually "I installed a binary from a source which carefully cloned the identity of a well-known source".
English
@flaviocopes TLDR; “I installed an unknown binary from the unknown source”. 🤷♂️
English
@aaxsh18 @cgarciae88 Nope, he can't eval the model, he can only eval your whole system. Which is very different from evaluating the embedding model.
English
@cgarciae88 hey you can upload the documents and then do your evals as normal. we are making the stores soon available for everyone to verify the evals. we also give reesearchers free credits to run benchmarks or try the system.
English
disappointing. how do you even benchmark this model?
Mixedbread@mixedbreadai
Introducing Mixedbread Wholembed v3, our new SOTA retrieval model across all modalities and 100+ languages. Wholembed v3 brings best-in-class search to text, audio, images, PDFs, videos... You can now get the best retrieval performance on your data, no matter its format.
English
@dylan522p You're more retarded than your take, and that's saying something github.com/openai/codex/i…
English
OpenAI Codex 5.4 + CLI sucks
Lazy as fuck and slow, it just uses my local files, tries to make up an answer from that rather than creating something
Doesn't have plugins like this
github.com/jarrodwatts/cl…
No /statusline to auto config my status bar either #cost-and-duration-tracking" target="_blank" rel="nofollow noopener">code.claude.com/docs/en/status…
English
@jasondeanlee @_onionesque And you got sacked from GDM, right? So you're very clearly biased.
English
@adamdotdev What would you know about adults in the room? You're clearly never gone beyond puberty.
English
@nicolaygerold Our high-speed trains are more on time than yours, arschloch
English
Another country that will soon be saying "we are late because of a delay in our neighboring country."
English
@xeophon Yes, but they want to define the models' values", or what they should answer in different cases. I don't think that their job is to define what is ethical, for example.
English
Anthropic's models are impressive, and I’m using them all the time, but publishing a 'constitution' isn't regulation. The gap between their frontier capabilities and self-imposed ethics theater is amazing. We need external AI regulation, not internal virtue signaling.
Anthropic@AnthropicAI
We’re publishing a new constitution for Claude. The constitution is a detailed description of our vision for Claude’s behavior and values. It’s written primarily for Claude, and used directly in our training process. anthropic.com/news/claude-ne…
English
Many SOTA models have over a trillion parameters, but this only only has 10 Billion AAAAAND its open source.
Hmm.....
MiniMax (official)@MiniMax_AI
MiniMax M2.1 is officially live🚀 Built for real-world coding and AI-native organizations — from vibe builds to serious workflows. A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench, exceeding leading closed-source models like Gemini 3 Pro and Claude 4.5 Sonnet. The most powerful OSS model for the agentic era is here.
English
Sometimes there are such horrors revealed on X that the only sanity-preserving move is to ignore them and pretend it's not happening
English
@Maggie0shaw @synthwavedd @mahaoo_ASI Not knowing shit is literally @synthwavedd trademark, together with his single digit IQ
English
@synthwavedd @mahaoo_ASI You don’t know shit. Just hyping for clicks like every other mentally ill AI influencer
English
Most interesting chart in this for me so far is this one:
I'm surprised about the the large MiniMax M2 share--this is not a model I hear much about.
OpenRouter@OpenRouter
We collaborated with @a16z to publish the **State of AI** - an empirical report on how LLMs have been used on OpenRouter. After analyzing more than 100 trillion tokens across hundreds of models and 3+ million users (excluding 3rd party) from the last year, we have a lot of insights to share.
English