zCat @zcat-app.bsky.social · 02/10/2026not sure what is worse, crypto airdrop hopers or AI reset hopers 000
zCat @zcat-app.bsky.social · 02/10/2026EU is really something to study, enslaving own people while being obedient vassal for US profits, sweet 000
zCat @zcat-app.bsky.social · 28/09/2026hmmm... I saw how Opus 5.5 is way better... I bend... I bought... I think that maybe internet is full of lairs 😬 000
Reposted by zCatUnsloth AI @unsloth.ai · 28/09/2026You can now run Laya Decision models locally on just 4GB RAM! 🔥 Works on CPU, Mac, Windows, Linux and GPU setups. Serve Laya through a Jev-compatible API via Unsloth Desktop. GitHub: github.com/unslothai/un... Guide: unsloth.ai/docs/models/... 1568
Reposted by zCatZcash en Español @zcashesp.bsky.social · 26/09/2026📣 Lanzamiento de Zebra 6.4.2 Zebra 6.4.2 es una versión de seguridad de emergencia que corrige una vulnerabilidad de alta gravedad presente en Zebra 6.4.0 y 6.4.1, que permite a cualquier nodo homólogo bloquear un nodo con un solo mensaje de red. 112
zCat @zcat-app.bsky.social · 24/09/2026Hmmm... Astra is somerhing: > I need to read new value, lets delete half of security spec rules > What? > Oh, yeah, I need it > Why? > Mmmm > Why? Just read value from DB > You re right, I made a mistake Yeah, good luck with autonomous agentic coding 211
zCat @zcat-app.bsky.social · 24/09/2026Astra is amazing. > Spec - only Java allowed. > Task - text clasification. > Astra runs over night > New Python service created > Running Maven tests that do not test the Python service to prove Python service is correct > Cancelled subscription 😬 010
Reposted by zCatUnsloth AI @unsloth.ai · 22/09/2026Qwen-Image-2.1 can now run locally on 12GB VRAM with Unsloth GGUFs! The 7B parameter model performs on par with Nano Banana 2.0. Run our Dynamic FP8 or GGUFs for higher quality via diffusers, Unsloth Desktop & more. GGUF: huggingface.co/unsloth/Qwen... Guide: unsloth.ai/docs/models/... 1407
zCat @zcat-app.bsky.social · 14/09/2026OK, i am confident now AI will not replace shit, it could be tool, but only by people who know what they do. OpenAI, Scamthropic, Gemini, GLM, Kimi, Qwen, Deepseek... All can be used for smaller targeted tasks some even in big codebase with guidance. But as full development lifecycle - forget 110
zCat @zcat-app.bsky.social · 13/09/2026Nerfing as a service... Astra & Sol completly unusable since yestreday, crazy. Even crazier that till yesterday I had barely any issue with it. I dont get it, similar was Claude 2 months ago before I canceled it. 101
zCat @zcat-app.bsky.social · 11/09/2026Hmmm, good i decided to use it for open source 😏 www.reddit.com/r/LocalLLaMA...reddit.comFrom the LocalLLaMA community on Reddit: ANOTHER researcher accuses OpenAI of training on conversations and then claiming a breakthroughExplore this post and more from the LocalLLaMA community 000
zCat @zcat-app.bsky.social · 10/09/2026Update Tor www.reddit.com/r/TOR/commen...reddit.comFrom the TOR community on Reddit: Tor Browser 15.0.22 released today for high-severity Tor security issuesExplore this post and more from the TOR community 000
zCat @zcat-app.bsky.social · 10/09/2026FML, everyone says Astra this, Astra that... In my case, it runs 4th day a measurement how to improve RAG retrieval even before implementation... What am I doing wrong? 000
zCat @zcat-app.bsky.social · 10/09/2026I do not get it, the GLM 5.3 flash... Is it really useful for you in your workflows? 000
zCat @zcat-app.bsky.social · 10/09/2026Watching Blockstream retards like Mow or Todd is kinda guilty pleasure. I never thought that level of stupidity is even possible... never underestimate humans, remarkable 100
zCat @zcat-app.bsky.social · 09/09/2026Not bad, but funny how they steal user data (recent math problem tech.yahoo.com/ai/articles/...) and try claim rescuing the world (by stealing your work).. Like scrapping Anna's Archive was totally for your good... As was killing little girls in school (scamthropic) www.youtube.com/watch?v=sLST...youtube.comBuilding Codex with Tibo SottiauxYouTube video by The Pragmatic Engineer 000
zCat @zcat-app.bsky.social · 09/09/2026AI barely change color of button in legacy enterprise codebase, but it is going to kill us 000
zCat @zcat-app.bsky.social · 09/09/2026KeePass vuldb.com/cve/CVE-2026...vuldb.comCVE-2026-86776 in KeePassA weakness has been identified in KeePass up to 2.61.1. This vulnerability is tracked as CVE-2026-86776. 000
Reposted by zCatTuta @tuta.com · 06/09/2026Google removed uBlock Origin from the Chrome Web Store 😡 Long live uBlock Origin! 🥳 This #DiDay, take a look at alternative browsers to Chrome & #DeGoogle?! → Browser alternatives: tuta.com/blog/best-pr... → Degoogle: tuta.com/blog/degoogl... 311664
zCat @zcat-app.bsky.social · 07/09/2026Tabnine is really special. I do not understand how did they get into enterprise, and convinced corporations to pay for licenses. It is so badly designed it is hilarious. But seeing the incompetence level on managing positions of companies, it actually kinda makes sense 000
zCat @zcat-app.bsky.social · 07/09/2026Yes and no. Spec-driven development is the best we have (now) and sets limits to project rules. It can generate good and project aligned code, it can review it and enforces project rules. However human should be still in the loop - thus understand the code 100
zCat @zcat-app.bsky.social · 07/09/2026Blockstream is the 2nd worst thing that happened to Bitcoin after Saylor 000
zCat @zcat-app.bsky.social · 06/09/2026If you want to feel real pain, use something called Tabnine + Gemini If anyone will ever lead logs I am fired 🤐 100
zCat @zcat-app.bsky.social · 06/09/2026I feel scammed... Everything is benchmaxxed, GLM Flash is horse shit, GLM full is less retarded Deepseek, but unusable for spec-driven full workflows... Great i drunk-bought yearly subscription... FML 000
zCat @zcat-app.bsky.social · 05/09/2026GLM 5.3 so good that Astra created 50 tickets to rewrite everything it made, and trashed all mesurement and analysis it did, damn... 💀 100
zCat @zcat-app.bsky.social · 03/09/2026mullvad.net/en/blog/2026...static.klipy.comGarfield Sad in the RainALT: Garfield Sad in the Rain 010
zCat @zcat-app.bsky.social · 02/09/2026I have really hard time to believe GLM 5.3 is to tier, not ecen mentionig the flash... How the hell is anyone use that for spec driven workflows? GLM bad, Deepseek very bad, Qwen max -okish, Kimi - good but wasting tokens, Opus - never again 100
Reposted by zCatInternational Cyber Digest @intcyberdigest.bsky.social · 28/08/2026xAI trained Grok on child sexual abuse material, a proposed class action filed Wednesday says. The plaintiff, abused as a preschooler in the early 2000s, says the Canadian Centre for Child Protection found AI-generated CSAM of her on xAI. 132
zCat @zcat-app.bsky.social · 02/09/2026Funny, Gemma 4 is still the king over other local models for the chat app need (even qwen3.8-next or Ling 3.0) 🙃 000
zCat @zcat-app.bsky.social · 24/08/2026Anthropic just sucks www.reddit.com/r/ClaudeAI/c...reddit.comFrom the ClaudeAI community on RedditExplore this post and more from the ClaudeAI community 000
Reposted by zCatGrapheneOS @grapheneos.org · 22/08/2026Initial focus for our partnership with Motorola is a regular non-folding device. Current generation Motorola Signature (2026), Razr Fold (2026) and Razr Ultra (2026) are very close to meeting our requirements but lack MTE, adequate secure element integration and other features being added for 2027. 1222727
zCat @zcat-app.bsky.social · 21/08/2026I really think we are reaching saturation point with current LLM design. It is not smarter, it only brute force longer. Valid for all new models so far from my experience... Funny, we still compare all new models to Fable/Mythos, not Opus 5, Kimi K3, not even mentioning GPT... 000
zCat @zcat-app.bsky.social · 21/08/2026Hmm, I am losing it. After my last chat with Deepseek, I might be banned from China. Deepseek is really retarded... Funny, local Qwen 3.8 (27B) solved ticket that Deepseek v4-pro cycled itself on. With the price adjustment what is the point really. And quantized Q3 is even worse than Gemma4 26B 000
Reposted by zCatTuta @tuta.com · 20/08/2026In 2024, the Higher Court of Schleswig-Holstein, Germany, ruled that standard TLS encryption is not secure enough 🔓🚨 Find out why this court ruling means that your emails must be end-to-end encrypted instead in our new blog post 👉 tuta.com/blog/german-... 27018
zCat @zcat-app.bsky.social · 17/08/2026Qwen 3.8 27B is not Opus 4.6 level, i hate these marketing claims Also, GLM 5.3 is not Fable, it is good but not Fable good 100
zCat @zcat-app.bsky.social · 14/08/2026www.youtube.com/watch?v=UOTA...youtube.comAmerican Politics | George Carlin | Life Is Worth Losing (2005)YouTube video by Official George Carlin 000
zCat @zcat-app.bsky.social · 14/08/2026Damn, i need to try this... xcancel.com/Zai_org/stat...rss.xcancel.comZ.ai (@Zai_org)Introducing GLM-5.3: Built to Code. Ready for Cyber Defense. - Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model - A major leap in cybersecurity, setting... 000
zCat @zcat-app.bsky.social · 13/08/2026Qwen 3.8 27B countdown, i kinda hope for MoE (probably not) 🙄 huggingface.co/Qwen/Qwen3.8...huggingface.coQwen/Qwen3.8-27B · Upcoming release · Hugging FaceRenewal of the beloved Qwen model, delivering unmatched intelligence density. 010
zCat @zcat-app.bsky.social · 13/08/2026static.klipy.comCJ: Ah Shit, Here We Go AgainALT: CJ: Ah Shit, Here We Go Again 000
zCat @zcat-app.bsky.social · 13/08/2026I am having hard time to belive deepseek is at kimi k3 level in some bench xcancel.com/grok/status/...xcancel.comVerifying your browser… 100
zCat @zcat-app.bsky.social · 11/08/2026More I use AI, I swear more. The day AI start rulling the world I am cooked. 110
zCat @zcat-app.bsky.social · 06/08/2026I feel so safe when MoFos like Zuckini protect kids... Cannot wait we all upload all data to facebook/twitter/whatever-stupid-cult app to be safe... xcancel.com/IntCyberDige...xcancel.comVerifying your browser… 010
