beowulf

🔊Introducing Voxtral TTS: our new frontier open-weight model for natural, expressive, and ultra-fast text-to-speech 🎭Realistic, emotionally expressive speech. 🌍Supports 9 languages and accurately captures diverse dialects. ⚡Very low latency for time-to-first-audio. 🔄Easily adaptable to new voices

I think that is the best advertisement I’ve ever seen.

📢 Introducing the AI Now Summit, Mistral AI’s first-ever flagship event! 🎯 One day, one mission: Own your AI transformation. 📍 Paris | May 28 Join us to learn how AI is transforming leading organisations and hear from global CEOs and Mistral’s founders on: ✅ Using open source as a core for end-to-end AI transformations ✅ Scaling from pilots to production deployments ✅ Building AI infrastructure for enterprise-grade deployments ✅ The latest in robotics, VLMs, and multimodal AI Learn more and get notified when tickets go live → ainowsummit.com

@FakePsyho The one about Poland looks very interesting I myself have worked with folks from Poland before and heard one of the main reasons is because the education system is very engineering focus, which is super nice

Testing interest in a few article ideas that I have (explanation of the ideas in the comments)

@FakePsyho I prefer personal blog post tbh I don’t want to be attached to this platform… there are days that I feel Twitter/X is getting worse Having a blog post allow you to control how you expose your content

My new year's resolution for 2026 is that I want to do more (useful) writing. Shitposting feels like gamified emptiness to me. What's the best way to do it? Preferably I'd just do a regular long post here, but sometimes you need pics along the way. Any feedback highly welcomed.

Un an après le premier Sommet de Paris pour l’action sur l’Intelligence artificielle, j’ai posé une question à Le Chat, notre pépite française par MistralAI, puis à ChatGPT.

Introducing Mistral AI's biggest hackathon ever! 📅 Feb 28 - Mar 1 🌍 Paris | London | NY | SF | Tokyo | Singapore | Sydney & online 48 hours. The best hackers. 🤝 Partners: @wandb @nvidia @awscloud @HackIterate 🏆 $200K in prizes. Special awards from @elevenlabs @huggingface @JUmp @whitecircle @supercell Link in 🧵

Introducing Voxtral Transcribe 2, next-gen speech-to-text models by @MistralAI. State-of-the-art transcription, speaker diarization, sub-200ms real-time latency. Details in 🧵

📣 Open call to agent builders: Let's read agent skills from `.agents/skills`, so people don't have to manage separate folders per agent. Today we pulled the trigger for Codex to read `.agents/skills`. Goal is to deprecate `.codex/skills`. Pls like/tag/RT for momentum.

got 1260 cycles, don’t think I can’t do better

The whole @AnthropicAI kernel challenge is very fun I did succeed to beat Opus, super happy but this shit was hard lol Got 1,357 cycles, now trying to optimise it more

@AnthropicAI leaderboard here: kerneloptimization.fun

Fantastic overview of all the ways you can sandbox untrusted code. Super thorough!

Great read about error handling in Rust but also in general for any type of services: fast.github.io/blog/stop-forw…

Came across this great blog from @beowolx on the foundations of sandboxes for AI workloads. Highly recommended if you're building AI agents luiscardoso.dev/blog/sandboxes…

@maximelabonne congrats mate, this is really a great model

Good night, sweet prince. It's been an amazing run. 🫡

@simonw Link here: luiscardoso.dev/blog/sandboxes…

My blog post received a nice shout-out from @simonw I'm super happy with the feedback so far! :) I hope it helps people navigate the subject, I do think this will be a super important thing this year.

I vibecoded an open-source poker river solver over the holiday break. The code is 100% written by Codex, and I also made a version with Claude Code to compare. Overall these tools allowed me to iterate much faster in a domain I know well. But I also felt I couldn't fully trust them. They'd make mistakes and encounter bugs, but rather than acknowledging it they'd often think it wasn't a big deal or, on occasion, just straight up try to gaslight me into thinking nothing is wrong. In one memorable debugging session with Claude Code I asked it, as a sanity check, what the expected value would be of an "always fold" strategy when the player has $100 in the pot. It told me that according to its algorithm, the EV was -$93. When I pointed out how strange that was, hoping it would realize on its own that there's a bug, it reassured me that $93 was close to $100 so it was probably fine. (Once I prompted it to specifically consider blockers as a potential issue, it acknowledged that the algorithm indeed wasn't accounting for them properly.) Codex was not much better on this, and ran into its own set of (interestingly) distinct bugs and algorithmic mistakes that I had to carefully work through. Fortunately, I was able to work through these because I'm an expert on poker solvers, but I don't think there are many other people that could have succeeded at making this solver by using AI coding tools. The most frustrating experience was making a GUI. After a dozen back-and-forths, neither Codex nor Claude Code were able to make the frontend I requested, though Claude Code's was at least prettier. I'm inexperienced at frontend, so perhaps what I was asking for simply wasn't possible, but if that was the case then I wish they would have *told* me it was difficult or impossible instead of repeatedly making broken implementations or things I didn't request. It highlighted to me how there's still a big difference between working with a human teammate and working with an AI. After the initial implementations were complete and debugged, I asked Codex and Claude Code to create optimized C++ versions. On this, Codex did surprisingly well. Its C++ version was 6x faster than Claude Code's (even after multiple iterations of prompting for further optimizations). Codex's optimizations still weren't as good as what I could make, but then again I spent 6 years of PhD making poker bots. Overall, I thought Codex did an impressive job on this. My final request was asking the AIs if they could come up with novel algorithms that could solve NLTH rivers even faster. Neither succeeded at this, which was not surprising. LLMs are getting better quickly, but developing novel algorithms for this sort of thing is a months-long research project for a human expert. LLMs aren't at that level yet.

Also go check what @e2b is building! Their product is super cool and it's definitely what I'd use

Here is the link: luiscardoso.dev/blog/sandboxes…

So, I spent the past couple of weeks working on a new blog post about "AI sandboxes" and sandboxes/VMs in general A practical guide to sandboxing for AI agents: how to reason about boundary vs policy vs lifecycle, where containers fail, and when to use gVisor / microVMs / Wasm / isolates.