Digging deeper on GLM-5 (dub.sh/glm5):

This 744B MoE beast (~40B active params) isn’t just “less woke” - it’s: COMPETITIVE.

Who’s running the GLM5 already?

Drop experiences in the comments, I'd like to exchange thoughts on it. 👇
#GLM5 #BasedAI #OpenSourceAI #LocalLLM #UncensoredAI #AgenticAI

Unlike American models (trained with progressive #RLHF in San Francisco),

China’s open-weight GLM-5 (dub.sh/glm5) is LESS “WOKE” (the community often calls it “based” or a breath of fresh air.)

🧵1/6
#GLM5 #BasedAI #OpenSourceAI #LLM #WokeAI

Tired of paying per token every time you autocomplete a function?
GLM-5 — runs local, integrates with anything.

Works with: Ollama · Continuedev · LM Studio · llama.cpp · OpenWebUI · AnythingLLM
→ dub.sh/glm5

#LocalAI #OpenSource #GLM5 #Ollama #AnythingLLM #LLM #DevTools #MLOps #CodingAI #FreeAI

Claude Code is impressive.
GLM-5 hits 77.8 on SWE-bench.

MIT licensed, open = run it locally, modify it, ship it.
→ dub.sh/glm5

#LocalAI
#ClaudeCode #CodingAI #OpenSource #GLM5 #ZhipuAI #LLM #AITools #SWEbench #MoE #AIAgents #DevTools #OpenWeights #CodeWithAI #AIEngineering #MachineLearning

GLM‑5 is released under the MIT license, with open weights, which lets anyone use, modify, and commercially deploy it with minimal legal friction.

self-deployment requires ~1,490GB of memory in native BF16 format

→ dub.sh/glm5

#GLM5 #OpenWeights #OpenSource #ZhipuAI #AIModels #LLM #MIT

🧵 The open-source model nobody talks about enough just dropped a technical paper last week, and the numbers are wild.

📊 #1 open model on SWE-bench Verified (77.8%) and Terminal Bench 2.0 (56.2%) — beating Gemini 3 Pro across the board.

Try it → dub.sh/glm5

#AI #OpenSource #LLM #GLM5

¡Adiós al "Vibe Coding", hola Agentic Engineering! 🚀
GLM-5 llega con 744B de parámetros para dominar el mundo Open Source. 🤖💻
✅ #1 en benchmarks de agentes. ✅ Tecnología DeepSeek DSA. ✅ Supera a GLM-4.7 y acecha a Claude 4.5.
youtu.be/hW6sqLlE7iQ
#GLM5 #AI #OpenSource

GLM-5: dilatih dengan 28.5 trilion token. Asal data? Tidak didedahkan.
Ini bukan masalah China semata-mata — ia masalah seluruh industri. Tiada model besar yang menerbitkan senarai sumber data lengkap.

Ketahui lebih lanjut: dub.sh/glm5

#GLM5 #KecerdasanBuatan #OpenSource #BeritaAI #LLM

GLM-5: addestrato su 28,5 trilioni di token.
Non è un problema cinese, è un problema dell'intero settore. Nessun grande modello pubblica un datasheet completo. "Open-source" significa pesi aperti, non trasparenza sui dati.
Leggi: dub.sh/glm5

#GLM5 #IntelligenzaArtificiale #OpenSource #AINews #LLM

GLM-5 : entraîné sur 28,5 billions de tokens. Origine des données ? Non divulguée.
Ce n'est pas un problème chinois, c'est celui de toute l'industrie.
Aucun grand modèle ne publie une fiche de données complète.
En savoir plus : dub.sh/glm5

#GLM5 #IntelligenceArtificielle #OpenSource #AIActualité

GLM-5 wurde auf 28,5 Billionen Tokens trainiert.
Das ist kein chinesisches Problem GPT, Claude, Gemini: alle gleich. "Open-source" heißt offene Gewichte, nicht offene Rechenschaftspflicht.
Details: dub.sh/glm5
#GLM5 #KünstlicheIntelligenz #OpenSource #Datenschutz #LLM

GLM-5: 28,5 biljoen trainingstokens. Databron? Niet vrijgegeven.
Dat is geen uitzondering — dat is de standaard in AI. Geen enkel groot model publiceert een volledig datasheet. Open-source betekent open gewichten, niet open verantwoording.
🔗 dub.sh/glm5
#GLM5 #KunstmatigeIntelligentie #OpenSource

Production runs on #OpenAI, #Anthropic, or a fine-tuned #Llama with a SLA behind it.

#GLM-5 is genuinely good,
dub.sh/glm5, but "good on paper" and "good in prod" are still two different products.

#GLM5 #LLM #MLOps #AINews #BuildInPublic #AICoding #AIRevolution #Coding #Programming #DevTools #AI

Zhipu AI is now Zai.
One letter. Global-friendly.
GLM-5 is a great model.
(dub.sh/glm5)

But the rebrand is its own strategy - and it's aimed squarely at Western developer adoption.

#GLM5 #ZAI #OpenSource #AIModels #LLM #GenerativeAI #AINews #MachineLearning #ZhipuAI #OpenWeights

Companies used to hire junior devs to do the grunt work while seniors architected.

GLM-5 (dub.sh/glm5) just got very good at the GRUNT work. The junior dev job isn't being disrupted. It's being deleted.

#AI #GLM5 #SoftwareEngineering #JuniorDev #FutureOfWork #Automation #DevCareer #TechTrends

The AGI frenzy is in full swing.
A quick reality check on GLM-5 (dub.sh/glm5): it hit 77.8 on SWE-bench. Solid score, no question.

That said, it’s self-reported on a test literally built for models to pass.

#AGI #GLM5 #SWEbench #AIHype #AICoding

China's GLM-5 drops with 744B params, 28.5T training tokens, and SWE-bench scores that beat Gemini 3 Pro on coding tasks.

dub.sh/glm5

All self-reported benchmarks, of course. The trajectory is hard to ignore. Open-source is where the real race is happening
#GLM5, #OpenSourceAI, #AITechRace, #GenAI

z.AI's #GLM5 (744B-A40B) is the biggest flex. At launch it benchmarked alongside GPT-5.2, Gemini Pro 3, and Claude Opus. Demand spiked so hard they raised prices on their coding plan: dub.sh/glm5

GLM-FileManager.py GitHub Gist: instantly share code, notes, and snippets.

Lil something I've been working on

GLM-5 File Manager Panel v1.1 for z.ai

#AI #GLM5 #VTSTech

gist.github.com/VTSTech/a560...

Claude Opus 4.6 costs $5 input / $25 output per 1M tokens.
GLM-5 ~$1 input / $3.20 output.

That's up to 8x cheaper.

🛠️ Plug it into Claude Code, Cline, Kilo, Cursor & more

📦 Plans start at $10/mo (Pro from $15/mo)

🎁 Promo code: MCERSXR9BE

🔗 dub.sh/glm5

#GLM5 #AI #DevTools #CodingA

I Use Claude Code But Not for My Personal Projects — Here's What I Use Instead Think of it as an AI agent that lives on your machine. Not a chatbot that gives you code suggestions. An actual agent that can create files, edit code, run commands, debug errors, and work through mul...

firethering.com/goose-open-s...

Finally a Powerful #opensource alternative to Claude Code That can run Online LLMs as well as Connect with Local OLLAMA LLMs. #chatgpt #Claudecode #glm5

Completely free & open source

China's GLM-5 Rivals GPT-5.2 on Zero Nvidia Silicon Zhipu AI's 744B open-source model GLM-5 was trained entirely on Huawei Ascend chips and now competes with GPT-5.2 and Claude Opus on major benchmarks.

China's GLM-5 Rivals GPT-5.2 on Zero Nvidia Silicon

awesomeagents.ai/news/glm-5-china-frontie...

#Glm5 #ZhipuAi #Huawei

GLM-5 now available in ai.to.design - ‹div›RIOTS Generate designs from prompts with GLM-5 in ai.to.design.

🚀 #GLM5 just dropped and it's already live in ai.to.design!

Z.ai's most powerful model brings cleaner, more sophisticated designs straight from a prompt — and the difference vs GLM-4.7 is very visible.
Try it now 👇
divriots.com/blog/glm5-n...

GLM-5: The World's Strongest Open-Source LLM Solely Trained on Chinese Huawei Chips The listed Chinese AI company Zhipu AI has introduced GLM-5, the fifth generation of its large language model. The model comprises 745 billion parameters

GLM-5 is een krachtig model, open weights en volledig getraind op Huawei Ascend-chips, zonder gebruik te maken van NVIDIA-hardware. Onderstreept het belang van Europese investeringen in AI.
www.trendingtopics.eu/glm-5-the-wo...
#ai #glm5 #huawei #nvidia #china

I hope all that information helps you get up to speed a bit.
For yourself as a teacher, you don't need to go this far really.
Check out the newthe beginning of #GLM5 where he creates a class in min.
Yes you can fine tune etc. Crazy awesome. I suggest you take this seriously
youtu.be/x9e4D-t_Dfo?...

Z.ai - Free AI Chatbot & Agent powered by GLM-5 & GLM-4.7 Meet Z.ai, your free AI-powered assistant. Build websites, create slides, analyze data, and get instant answers. Fast, smart, and reliable, powered by GLM-5.

Chinese players like #Zhipu, #Moonshot & #MiniMax have yet to go public. We see Z.ai, the makers of #GLM5, shot up 34% when releasing it on the HK market. Yes co get smaller injections of capital funding rounds of only $500 mln unlike the recent $30 bln injection into #Anthropic, $380 bln eval.

This is annoying. #llm #zai #dev #opencode #mcp #glm5

I switched to PRO plan on my z.ai coding subscription to try out faster #glm5. So my first comment, it is much slower then 4.7, but overall all z.ai infrastructure today is kind of underperforming - including web interface.

Let me try it more and I will […]

[Original post on mastodon.com.pl]

Another very strong endorsement by The O t3dotgg on #YT of #GLM5 by z.ai. Theo is #dev, YouTuber & also a founder of YCombinator.
Very hard to get more respected in the #AI VC space. He of course worked w/ Sam Altman at YCombinator.
"The O" is a term for Operating Partner.
youtu.be/vtWMgVCMsx8?...

✨ GLM-5, il nuovo modello di Z.AI si avvicina a Claude Opus 4.6 nei benchmark di coding.
Architettura MoE da 744B parametri e un costo di soli 0,80 dollari per milione di token.

gomoot.com/glm-5-il-nuo...

#ai #glm5 #news #tech