Obsessed by automation. Unstoppable. I tame and teach machines 🇪🇺🇮🇹 "Through the fire, and the fury, with a heart made of steel". OSS/Acc.
Milan, Italy
Joined May 2011
- Tweets13.8K
- Following627
- Followers1.8K
- Likes38.2K
Jacopo Nardiello retweeted
Mistral Large 4, all 1 trillion parameters of it (3-bit EXL3 quant), now runs on 4x DGX Sparks 🐈
• up to 40 tok/s decode
• ~600 tok/s prefill
• 65K context
• one command to install
Fitting this chonk onto four small boxes was a squeeze.
github.com/grubbychicken/mis…
Jacopo Nardiello retweeted
🔴 CRITICAL 9.5 · React: remote code execution in server components (CVE-2026-126187)
An unauthenticated attacker can run arbitrary code on the server by sending a crafted request, without logging in.
👉 Fix: 19.0.9, 19.1.10 and more
cve.now/CVE-2026-126187?utm_…
Jacopo Nardiello retweeted
This is a good filter for people who I don’t want to work with. Please understand how computers work. This isn’t even about understanding Docker. Any software person should know how the layers work at a high level and this is a VERYYYYY high level.
Yes, and this isn't accidental.
They are all compute-constrained. When a new model is announced, they allocate compute to it so it shines and users re-subscribe, while quantizing and nerfing the old models.
Thank you for attending my TED Talk.
Anyone notice how every time there's a new model the last one loses half it's braincells.
Anyway, Anthropic is on a biblical run
3
145
Let’s hope Opus 5.5 will also be back
BTW, Opus got nerfed to the ground - what did they do to my boy?
I consistently used it to manage end-to-end stuff at my print farm (yes, I have a small one). It could on its own:
- manage slicing software
- send jobs to printers
- handle all listings on eBay/Amazon/Shopify
Now it gets confused on basic stuff. Not joking, but performance is comparable to my local GLM 5.3 Flash (at least on computer use). Bad @AnthropicAI, bad!
Now give us Fable 5.5 please!
109
Jacopo Nardiello retweeted
Introducing Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges.
They’re capable collaborators on ambitious projects in many fields, and show improvements in areas like coding, reasoning, and vision.
BTW, Opus got nerfed to the ground - what did they do to my boy?
I consistently used it to manage end-to-end stuff at my print farm (yes, I have a small one). It could on its own:
- manage slicing software
- send jobs to printers
- handle all listings on eBay/Amazon/Shopify
Now it gets confused on basic stuff. Not joking, but performance is comparable to my local GLM 5.3 Flash (at least on computer use). Bad @AnthropicAI, bad!
Now give us Fable 5.5 please!
Jacopo Nardiello retweeted
Breaking News: Nvidia opened an investigation into Jensen Huang, its CEO, after he acknowledged to the company’s board that he had been intimate with a business executive who U.S. intelligence determined to be a Chinese spy. nyti.ms/3ViPMAt
Jacopo Nardiello retweeted
huggingface.co/ukisai/Swift-… Trained to not overthink, good daily driver
huggingface.co/cantina-secur… trained for cyber work
There's many more post trained glm ones unlike deepseek, only tested these tho
Jacopo Nardiello retweeted
Kimi K3.1 Pro open-weights are out.
Weights ~1.5TB
huggingface.co/moonshotai/Ki…
Jacopo Nardiello retweeted
SITUATION DETECTED: The EU refrigeration cyberattacks have escalated to medical infrastructure. Hackers breached Coolworld, a French national blood bank contractor, changing transfusion plasma deep-freeze temps from -30°C to 110°C. It was caught before the blood was boiled.
Jacopo Nardiello retweeted
BREAKING: Revolut moved ~2,000 engineers onto self-hosted GLM for coding and pushed Claude Code and Codex to the exception list.
Per David Abitbol, who leads AI ops there: 56 Nvidia B200s, a blended 30 cents per million tokens, roughly a tenth of what they paid Anthropic and OpenAI. Engineering lead time improved 70%.
Here's the part that matters: the gains showed up when tokens stopped being rationed. Engineers quit saving AI for the hard tickets and pointed it at everything.
Frontier models become the exception, open weights become the default. That's a pricing problem for every lab heading to IPO.
h/t @OliverMolander
Would you move your team off frontier APIs if the bill dropped 90%?
