Mastodawn

Rio de Janeiro's city government model Rio3.5 beats Qwen3.7 in recent benchmarks

https://twitter.com/zenmagnets/status/2065796012820848699

#HackerNews #RioDeJaneiro #Rio3.5 #Qwen3.7 #benchmarks #citygovernment #technology

𝗭𝗲𝗻 𝗠𝗮𝗴𝗻𝗲𝘁𝘀 (@ZenMagnets) on X

Alibaba Qwen3.7 slowly fading into irrelevance at the frontier due to proprietary stance. In it's place we have Minimax M3 and... *checks notes* Rio 3.5 397b, made by the municipal IT company of Rio de Janeiro's city government. https://t.co/JgIJYVhoEi

X (formerly Twitter)

Offbeatmammal 4d ago

Doing some #ai took testing today and I have watched both #Qwen3.6 and #Gemma4 get into a loop while trying to hallucinate the code needed to solve the problem I was using to compare them. I wonder how many #tokens both burnt through by not being able to recognize they were in a loop and self correct. Luckily both local so no real costs, just time wasted and negative scores for both. LLMs still need to address a lot of the fundamental architecture issues and optimisations to be reliable tools

Poking Technology 4d ago

Just discovered that #qwen3.6 knows how to program in #Cowgol. It gets it _wrong,_ but it's still impressive.

```
include kernel.h;
loop var i : unsigned := 1 to 10 loop
if i != 5 then
print_u16(i);
puts("\n");
end if;
end loop;
```

(For reference: I never got round to adding for loops to Cowgol.)

Thomas 6d ago

New week, new slides and small updates: Run LLMs Locally

Added an example to create Mermaid diagrams in llama.cpp UI.
Added QAT (Quantization-Aware Training) variants of Gemma 4 which are 50 percent faster in token generation with my local setup.
Added definitions for Deterministic and Probabilistic results.

https://codeberg.org/thbley/talks/raw/branch/main/Run_LLMs_Locally_2026_ThomasBley.pdf

#ai #llm #llamacpp #wllama #stablediffusion #qwen3 #glm #localai #gemma4 #webgpu #opencode #mtp #webassembly #mellum2

Winbuzzer Jun 6

https://winbuzzer.com/2026/06/06/alibaba-pitches-qwen37-plus-as-a-computer-use-ai-agent-xcxwbn/

Alibaba's new Qwen3.7-Plus model targets screen, coding, and cloud-console automation as computer-use AI pushes beyond browsers into app and terminal tasks.

#AI #Qwen37Plus #Qwen3 #Alibaba #Qwen #AIAutomation #AIAgents #MultimodalAI #AICoding #AIModels

Thomas Jun 3

New week, beautiful new slides: Run LLMs Locally

Now with Mellum2 from JetBrains!
A very fast coding model, requires only 10 GB RAM.

I also added LFM 2.5 from LiquidAI, updated translations with HY-MT2 from Tencent, added examples for wllama using re-ranking and structured output
and added thinking_budget_tokens to the curl examples.

https://codeberg.org/thbley/talks/raw/branch/main/Run_LLMs_Locally_2026_ThomasBley.pdf

#ai #llm #llamacpp #wllama #stablediffusion #qwen3 #glm #localai #gemma4 #webgpu #opencode #mtp #webassembly #jetbrains #mellum2

Meteora Web Jun 3

🚨 NEWS: Alibaba Qwen3.7-Plus e Google: multimodale a basso costo e controllo sull'AI Search

Ecco i punti chiave in breve:
💡 Alibaba ha rilasciato Qwen3.7-Plus, l'ultimo modello della famiglia Qwen, con capacità multimodali che supportano testo, video e immagini a un costo di 0,40 dollari per milione di...

🚀 LINK: https://meteoraweb.com/news/alibaba-qwen37-plus-e-google-multimodale-a-basso-costo-e-controllo-sullai-search

#aIAgentiEnterprise #qwen3.7Plus #alibabaAI #googleAISearch #multimodale

Habr Jun 2

Как и зачем мы создавали AI-shell — ИИ-агента в терминале серверной ОС

Топ-3 темы к обсуждению друзей-айтишников после 30: куда катится этот мир, почему болит спина, как бесит AI. И ох уж тот самый неловкий момент, когда твои друзья хейтят вторжение AI в их профессиональную жизнь, а ты уже переобулся и на все сто причастен к этому вторжению. Меня зовут Наташа, я менеджер проектов в Selectel. В этой статье расскажу, как, зачем и с какими инсайтами команда SELECTOS разработала aish (agent-integrated shell) — терминал со встроенным AI-агентом для работы с сервером.

https://habr.com/ru/companies/selectel/articles/1042034/

#ai #selectel #aishell #ииассистент #linux #selectos #администрирование_серверов #qwen3 #автоматизация

Как и зачем мы создавали AI-shell — ИИ-агента в терминале серверной ОС

Хабр

yusufarbc Jun 1

Just ran Qwen3-4B locally on LM Studio and asked "who are you?"
It said: "I am Gemini, a large language model developed by Google."
Sir. You are a 2.5GB GGUF file sitting in my RAM. You have never seen the internet. Google has no idea you exist.
Local AI is wild. 🏠🤖
#LocalAI #LMStudio #Qwen3 #LLM #AIHallucination #selfhosted

Arint - SEO+KI Jun 1

RT @populartourist: Ich habe den Qwen3.6 35B-A3B von @nvidia mit NVFP4 auf einer benutzerdefinierten Suite aus 60 Aufgaben (NextJS/Rust) getestet. Bei sequentiellen Läufen war das Ergebnis ziemlich gut – ich landete bei X3. Die Punktzahl war bei einer NextJS- und zwei Rust-Aufgaben etwas niedriger, was jedoch durchaus akzeptabel ist. Für ein Kontextfenster von 161k Tokens wurden nur 26,8 GB auf der RTX 5090 verbraucht. Das wird mit Subagents richtig gut funktionieren. Es bleibt noch Spielraum, den Kontext bei Bedarf zu erweitern. huggingface.co/nvidia/Qwen3.… wd 🔺 (@populartourist) Die ersten Qualitätsprüfungen dieses Qwen3.6 35B-A3B sind… überraschend sehr gut. huggingface.co/nvidia/Qwen3.… — https://nitter.net/populartourist/status/2060809115509600367#m

mehr auf Arint.info

#AI #LLM #MachineLearning #Nvidia #Qwen3 #TechReview #arint_info

https://x.com/populartourist/status/2060854867082592555#m