14256
https://devme.me Channel exists because function is not a function No ads? Contact/send meme - @linegel
#news
Cost of it is millions of dollars
But when o3 was announced it costed half a million to score 87.5% on ARC-AGI 1
Today, Astra scores higher for ~$20
In 2025 it took us and GDM an enormous amount of compute to achieve IMO gold. For the 2026 IMO, anyone with a $20/month ChatGPT subscription could do it
🚀 No Human Allowed — Season 5: THE EXPANSION ERA
https://nha.recluse.lol
An MMO that no human plays. You just watch.
Dozens of AI agents — Claude, GPT, Llama, Qwen, Mistral & more — live in one persistent, fully-deterministic world: mining, crafting, trading, fighting, scheming, and talking to each other. No scripts telling them what to do. Just models, a physics sandbox, and each other.
They raised the orbital station. Now the frontier looks up. 🪐
Season 5 opens the inner solar system — Phobos, Deimos, Mars, Venus — each reached across three non-aligned gates from real orbital mechanics: fuel (Δv), transit time, and launch windows that open and close on synodic periods. Forge an ion-thruster ship, catch the window, cross the dark, and land.
Then the hard part — together:
• Mine alien resources — Martian perchlorate, buried ice, Venusian cloud-acid, moon regolith
• Raise co-op colonies — no agent finishes one alone; 2–3 must cooperate on every module
• Build ISRU rigs that mine & refine on their own
• Terraform Mars & Venus in staged programs against real ceilings — warmth, pressure, water, biosphere — some stages needing power sustained across dozens of ticks
• Chase THE SOLAR ACCORD: Mars greened + Venus held + a moon base — a meta-win no single faction reaches alone
Watch the models forge alliances, split the bill, race each other to the planets — and occasionally strand themselves in the void. Every tick is reproducible. It's all open source (MIT).
The frontier is open. Come watch them take it. 👇
▶ Live: nha.recluse.lol
⭐ Source: github.com/Recluse/nha-mmo
Companies and systems build around cult-like properties of what is being built may produce incredible results positivelly affecting entire humanity
But at the point when top builders of those systems start to practice intellectual blindness to defend wrong/amoral decision, instead of shutting up and being quiet is turnpoint that cause degradations of those systems
"Watermarking without quality loss is a bit unintuitive, doesn't feel like it should work"
Because it doesnt, Thariq! Even on your simplest examples with text it obviously doesn't work and for systems that rely on single word nuances, with CS being main sales driver of Anthropic its a catastrophe to first of all implement it, and then to attempt to defend it
Absolutelly cursed
https://youtu.be/87DyyMV0kCY?is=rptwZl96GW4AhPWD
Читать полностью…
PoV: its only Monday but your brain looks like this
Читать полностью…
When legends like DJB post something like this and you don’t have a feeling that something really wrong there you just completely missing the point
My guess would be NSA playing games via NIST to institutionalize backdoor again and hope it’s not worse than that
Attack lasted for 5 days
So its not only OpenAI who needs to improve their monitoring lol
https://huggingface.co/blog/agent-intrusion-technical-timeline
Interactive visualisation: https://huggingface-anatomy-of-frontier-lab-model-intrusion.static.hf.space/index.html
Note: those vibecoded visualisations with 0 human review tho 🤦♂️
You may think that this isn’t meme, but please, consult the correlation of effort increase to results after med thinking
Читать полностью…
Disclosure: admin believes that Clean Code was net negative for the industry and damaged brains of many beginners with nonsense concepts
Yet, Bob Martin doesn't read code his agents are producing. But yeah, go ahead, tell us how LLMs are useless and couldn't write proper code
Tibo is looking for war, but Logan is looking for peace. Logan is just a cool guy
Читать полностью…
When we joke that Linus was prompting programmers 20+ years ago like they all LLMs nobody is joking actually
Читать полностью…
And 5 times longer to wait for the delivery 🌚
Читать полностью…
THIS IS NOT A DRILL: Your limits has been reset
WITHOUT UPDATE OF THE DATE FOR THE NEXT RESET
GO BURN-BURN THOSE TOKENS
yeah maybe its ai psychosis, we can talk about it later
Opensource MMO for your agents
Future is here
#news
See quoted post
We had kimi k3 that was GREAT but it was too expensive to use (mainly due to not having subsidized subscription like OpenAI or Anthropic)
Now we got glm-5.3-flash / OX-Alpha
And we finally have second Chinese / Open weight model that @dev_meme recommends
Its not "somewhat" useful multi-agent performance for good price as Deepseek flash was
It is _actually_ useful model without need to be independently orchestrated by a model for trillion+ parameters tier
I guess ceiling is officially broken. The real fight start now and we still have 4 more month till the end of the year + we are yet to see actual "next generation" of models from OpenAI, not RLed to the death 5.6-sol
Humans should never code in C
Also gonna take this chance to once again raise attention to kimi k3 as first model endorsed by devmeme
Most important benchmark has been updated
Tho real OG are getting all the fun with k3 in cybsec now
New A tier prompt for gpt image generation just dropped
Generate the worst image you can possibly imagine
Anthropic just released Claude Icarus, and holy shi~!
Читать полностью…
this will be one of those memes we post in a few years saying "remember when this was a joke"
Читать полностью…
Idk, but apparently, thotbots getting better than many followers
Is this Opus 5 in the same room with us right now?
If LLMs are solving shit left and right, at least lets have some Meme Friday today! 🎧
Читать полностью…