Alex Xander

Gazing into the future... puzzled. Critically looking at achievements in Robotics, autoregressive/diffusion/JEPA architectures with an AI/IT security background

[Alex Xander @naturalSmartnes]

@MiaAI_lab I'll try and be unbiased

[Alex Xander @naturalSmartnes]

@araseb_ Building what ... I need. Fully local, multi-project agent orchestrator among all providers. Supports my BDD workflows and ensures I can always work even if Anthropic bans me for no reason 😆

[Alex Xander @naturalSmartnes]

@bridgemindai go for it! curious what's your burn rate

[Alex Xander @naturalSmartnes]

@onusoz @tunahorse21 Quick question: is the RDMA interoperable? Can you build shared memory between say a Dell and an Asus?

[Alex Xander @naturalSmartnes]

@NVIDIARTXSpark @HP These OmniBooks can likely make coffee and fry eggs!

(coffee was a joke)

[Alex Xander @naturalSmartnes]

Anthropic and OpenAI add very nice and useful features hoping to lure and lock you in.

Use fully but don't depend on them.

[ClaudeDevs @ClaudeDevs]

We've just added two new Claude Managed Agents features:

1. Scheduled deployments - run tasks on a schedule

2. Environment variables - expose vault credentials for CLIs as environment variables

[Alex Xander @naturalSmartnes]

@Sauers_ The test is in French

[Alex Xander @naturalSmartnes]

@sabari_246 @claudeai I noticed the opposite frankly. Not token maxxing here, but still cleared 10 development tasks in the last hour on a Pro Plan and weekly usage 31 -> 44. Using Sonnet medium as orchestrator Fable medium as implementor and reviewer

[Alex Xander @naturalSmartnes]

@ylecun @ClementDelangue @Dan_Jeffries1 I listened to your podcast on Unsupervised and totally agree. Not sure how I can help... Multiparty computation?

[Alex Xander @naturalSmartnes]

@petergyang There used to be an Antigravity IDE. They nerfed it up 3 weeks ago (when it became like ChatGPT) and people rebelled and uninstalled the new version and Google did a U turn since and the old IDE is again available. https://t.co/gURnlQoFrp

[Google Antigravity @antigravity]

Our biggest Mission Report update ever!

[LeRobot @LeRobotHF]

VLA-JEPA just dropped in LeRobot 🤖

What makes this model special is that it does not just learn what action to take from a given observation, it also leverages a JEPA world model to learn action-relevant dynamics.

During training, the VLA leverages V-JEPA2 by conditioning its predictor. This clever trick adds a world modeling objective to the training, which also allows pretraining on human videos.
At inference, the world model is dropped entirely, keeping only a standard VLA architecture: Qwen backbone and action head.

The demo here was only fine-tuned on 13 examples, showing great pretraining capability and running in real time on @NVIDIARobotics DGX Spark!

VLA-JEPA is the first world model to be ported to LeRobot, and I feel like it won't be the last 🚀

@Thom_Wolf @ClementDelangue

[clem 🤗 @ClementDelangue]

Concentration of power, capabilities and economic wealth is the biggest risk in AI. We need open science and open-source more than ever!

[Alex Xander @naturalSmartnes]

At least the increase of hardware prices is linear

[Mia @MiaAI_lab]

🤖 Fable 5 costs 2× Opus 4.8
Opus 4.8 costs 2× Opus 4.7
Opus 4.7 costs 2× Sonnet

📈 The pattern is clear.

⏳ Where are we in 1 year? 2? 5?

⚠️ Don't get priced out.
🏠 Run LLMs locally, on your own hardware.

[Alex Xander @naturalSmartnes]

@MiaAI_lab @0xSero 40 tok/s is good! Could you give some hints on the level of capabilities of Deepseek v4 Flash (at FP8 if I understand correctly)?

[Alex Xander @naturalSmartnes]

I have no clue who Pratham is, but he's totally right

[Pratham @Prathkum]

Hot take:

We don't need a more powerful model like Mythos right now.

GPT-5.5 (5.6 coming soon), Opus 4.8, and similar models are already more than capable for most use cases.

What we need to solve is the cost problem.

If AI keeps getting significantly more expensive, 99% of developers won't be able to afford these models at scale and will end up going back to manual coding.

[Alex Xander @naturalSmartnes]

@cristinaibunea The problem is that to enter the market as a vendor, you'd need to compete with big names that provide highly integrated, decent-security options for free.
As a consumer I think and hope that the sandbox won't go away as a product.

[Alex Xander @naturalSmartnes]

Pity it's only on the Ultra plan. But frankly the value of the regular NotebookLM on a free subscription is just crazy

[NotebookLM @NotebookLM]

Want a closer look at today’s launch? Here is a breakdown of what’s new and exciting 🧵:

First up: An upgraded, more thoughtful chat experience.

Powered by Gemini 3.5 and @Antigravity, you will now have better visibility into the AI's thinking process. Plus, each notebook has a secure cloud computer including 100+ curated software skills, unlocking deeper research and more complex analysis.

[Alex Xander @naturalSmartnes]

a new perspective on /loop

If it was supported by Codex, Antigravity and Github Copilot it would be an industry standard as Cursor, Claude Code and Mistral Vibe support it. Pun intended