paoliniluis

You can use it to connect every single data source that's around your company and then use it as a federation engine (e.g. spreadsheets!)

Here's a (vibe coded) Metabase driver that speaks Quack: https://github.com/dragonsahead/duckdb-quack-metabase

The fact that can now be used as a client-server architecture is a game changer. It will for sure replace many data stacks that now run on many servers with tons of cpu and ram

I’ve been applying for jobs in the last couple of months and on 80% of the sites I needed to include my LinkedIn profile or I couldn’t send my resume… so I had to forcefully create a profile there so random people could spam me vibe coded ROI calculators via messages. Seems like the system might be rigged

Seems that Sacks didn’t get that the message was targeted to the president to ask for a bilateral agreement with China, that’s the end game

Hey this is neat and I just found out you're from BA, Argentina. I would love you be part of dataengine.ar and maybe talk about arc?

Would be awesome to see a new ds4 release. Having so much in something that can be run locally is mind blowing

Just use moonlight and sunshine

If this article was true then Cerebras would be worth at least 2.5T and Nvidia half of what’s worth now. The market seems to be paying a premium for training chips rather than inference. Oh, and there’s also a bubble as you can see in Shiller’s PE ratio here https://www.multpl.com/shiller-pe

Using AI slop in software that runs ML models is like buying a Ferrari with a motorbike engine. Might be good for demos or as a proof of concept, but that can’t be used for serving LLMs at scale

There's a guy trying to take down the server by sending as user/pass the lyrics of Rick Astley's "never gonna give you up"

The IPO was meant for the VCs to cash out as all fundamentals were completely irrational, but seems like no one cares about cash flow and profitability anymore during QE times. Dumb money will keep being dumb I guess

Legends

The FED will print to infinity as the US gov can’t stop spending, mostly all of that money will keep going to the only industry that’s growing and provides crazy returns for family offices and VC’s right now which is AI. I don’t agree with the authors opinion here as the “time bomb” timer is simply the entire world buying US debt here, which won’t happen in the short/medium term

Alibaba's cloud is something that the author of this article seems to dismiss. It's being used massively in Asia and they're pretty close in services and offerings to what AWS, GCP and Azure provides. Once they start doing inference on their own custom chips it might be hard to compete with them due to the energy costs

They claim they're surpassing the latest DeepSeek model in many tests. I'm wondering when we can see a GGUF release of this model as being a MoE this should run fine on local machines.

BTW: here's the blog post https://mimo.xiaomi.com/mimo-v2-5-pro. They state that "DeepSeek V4 Pro numbers are with its max effort setting." so I'm wondering what they used for this one.

Agree. Google can't go against the all-mighty state. Just look at what Anthropic did and the effect of that action. There are billions of dollars at stake on government contracts that they can't afford to lose. Reminds me of Mullvad's ad https://www.youtube.com/watch?v=fPzvUW8qaWY

It's 2026 and there's still people that believe that proof-of-work makes sense as a consensus mechanism

Does everyone agrees that this makes Dario Amodei more powerful than any politician across the world? Anthropic is now the owner of the most powerful cyberweapon ever made

Who wants to vibe code an open source Cloudflare?

LinkedIn feed now brings dumb posts from AI bots that contacts follow. All social networks tend to follow the same principles now: bring to everyone’s feed what’s most engaging, which is normally clickbaits or posts that use exaggerated words

run ollama show <name_of_model>:<parameters> and you'll get the info. E.g. ollama show qwen3.5:0.8b Model architecture qwen35 parameters 873.44M context length 262144 embedding length 1024 quantization Q8_0 requires 0.17.1

  Capabilities
    completion    
    vision        
    tools         
    thinking      

  Parameters
    presence_penalty    1.5     
    temperature         1       
    top_k               20      
    top_p               0.95    

  License
    Apache License               
    Version 2.0, January 2004

just finding the perfect spot between accuracy of the answers/available VRAM/tokens per second

what's your take between Qwen3.5-35B-A3B and Qwen3-Coder-Next?