@elonmusk Big if true. What most overlook is this accelerating private surveillance tools, by 2027 we could see mandatory browser plugins for โcompliance.โ You think that timeline is realistic or too alarmist?
Anthropic just dropped Claude Fable 5 โ their new Mythos-class model optimized for safe, general use. Itโs crushing it in software engineering and scientific research, beating competitors by a solid margin. Perfect for heavy coding, deep research, and complex knowledge work.
Introducing Claude Fable 5: a Mythos-class model that weโve made safe for general use.
Its capabilities exceed those of any model weโve ever made generally available.
@karpathy This is a super exciting release indeed. I've been dealing with models that lose the plot halfway through complex sessions, and this sounds like a direct fix. The part that stood out was that "it's never felt this tempting to stop looking at the code at all."
The Cloudflare team, alongside xAI, came to a recent developer showcase to explain how Grok models are now native to AI Gateway, including all modalities, text, audio, image, and video, via a single endpoint.
They said: "No additional auth or API keys needed, you just call the model through your existing Cloudflare account and it's billed directly."
We're partnering with @xai to bring Grok to @Cloudflare AI Gateway.
โข Grok LLMs, audio, image, and video models are now available through AI Gateway
โข Billed directly through Cloudflare
โข No additional auth, env, API keys needed
๐๐๐.๐ฐ๐ธ.๐๐๐("๐ก๐๐/๐๐๐๐-๐๐๐๐๐๐๐-๐๐๐๐๐-๐ท.๐ป-๐๐๐๐๐๐๐ ")
@ParikPatelCFA Bro already tried it, but the 'Ask Pepper' button on the chatbot is disabled. Now I can only see the burrito menu. Wanna try? Click here & hit Ask Pepper
https://t.co/3BHDYw6g8e
@UnslothAI 12B on 8GB RAM is wild. I've been dealing with memory bottlenecks for smaller models and this changes the game. The part that stood out was the 256K context support. Have you tested it on long documents yet?
@heynavtoor To anyone who's curious about the project, here's my demo. The main feature I'm currently testing is the agent itself โ how independently it runs with local development. Stay tuned!
https://t.co/8zijlCKAcn
PewDiePie just built Odysseus - a fully self-hosted, private AI workspace!
Like ChatGPT but runs on your own PC. Supports local models, agents, web search, notes, and more.
I tried it myself โ super clean.
Full quick demo ๐
Repo: https://t.co/X8kavmlni7
Would you use this instead of ChatGPT?
#PewDiePie #Odysseus #LocalAI
@CollinRugg Big bust. One thing most people miss is the shift to urban fronts over rural routes. For those starting in law enforcement, prioritize following suspicious business patterns not just known hot spots. What's your experience with this kind of advanced smuggling infrastructure?
@Microsoft Just read through the details. The good part: topological qubits promise way fewer errors than traditional ones. The honest part: we're still years from a full-scale quantum computer that outpaces classical ones.
@pelositracker This vs traditional insider monitoring for compliance, stock tracking wins on speed and public data, but manual cross-referencing is better for context.
@rkmtimes Big geopolitical play. One thing most people miss is how this reshapes Apple's control over hardware-security integration. For those starting out, follow Israeli startups in custom silicon. What's your experience with Apple's chip roadmap?
@ism_sol Bisa dipakai untuk bermacam-macam. Salah satunya feature yang saya sukai adalah dia bisa melakukan compare terhadap 2 model yang berbeda secara langsung.