Albert

Todo está inventado o será inventado pronto.

Claude Code: "You've hit your limit · resets 7pm" Me from 5-6.59pm

Introducing Claude Opus 4.6. Our smartest model got an upgrade. Opus 4.6 plans more carefully, sustains agentic tasks for longer, operates reliably in massive codebases, and catches its own mistakes. It’s also our first Opus-class model with 1M token context in beta.

+1 for "context engineering" over "prompt engineering". People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window with just the right information for the next step. Science because doing this right involves task descriptions and explanations, few shot examples, RAG, related (possibly multimodal) data, tools, state and history, compacting... Too little or of the wrong form and the LLM doesn't have the right context for optimal performance. Too much or too irrelevant and the LLM costs might go up and performance might come down. Doing this well is highly non-trivial. And art because of the guiding intuition around LLM psychology of people spirits. On top of context engineering itself, an LLM app has to: - break up problems just right into control flows - pack the context windows just right - dispatch calls to LLMs of the right kind and capability - handle generation-verification UIUX flows - a lot more - guardrails, security, evals, parallelism, prefetching, ... So context engineering is just one small piece of an emerging thick layer of non-trivial software that coordinates individual LLM calls (and a lot more) into full LLM apps. The term "ChatGPT wrapper" is tired and really, really wrong.

We estimate that, on our tasks, Claude Opus 4.5 has a 50%-time horizon of around 4 hrs 49 mins (95% confidence interval of 1 hr 49 mins to 20 hrs 25 mins). While we're still working through evaluations for other recent models, this is our highest published time horizon to date.

We’re opening applications for the next two rounds of the Anthropic Fellows Program, beginning in May and July 2026. We provide funding, compute, and direct mentorship to researchers and engineers to work on real safety and security projects for four months.

New on the Anthropic Engineering Blog: Our tips for developers on using Agent Skills, a new way to extend Claude's capabilities with instruction folders, scripts, and resources: anthropic.com/engineering/eq…

🌟 Wrapping up X4HPC Annual Conference & Innovation Journey Kick-Off 2025! 🌟 We celebrated the X4HPC network's Annual Conference at Pier07, Barcelona, uniting researchers and collaborators. An inspiring day of exchanges, ideas, and new connections! #HPC #AI #DeepTech 1/6

@EricBuess Try @WisprFlow 🗣️

One of the tradeoffs of the CLI. And another of the many reasons I’m building a voice-first interface for Claude Code. 🤣

We’ve made three new AI fluency courses, co-created with educators, to help teachers and students build practical, responsible AI skills. They’re available for free to any institution.

7 Layers of LLM Stack

Esta es la realidad sobre los programadores en la era de IA. Uno de los mensajes más claros y potentes que he escuchado: 1️⃣ El peor consejo profesional: Aconsejar a la gente que no aprenda a programar porque la IA lo automatizará es uno de los peores consejos profesionales que se pueden dar. 2️⃣ La tecnología facilita, no elimina: Históricamente, cuando la programación se ha vuelto más fácil (del lenguaje ensamblador a COBOL, de tarjetas perforadas a teclados), el número de programadores ha aumentado, no disminuido. La IA es una herramienta que hará la programación más accesible. 3️⃣ Demanda de software casi infinita: La necesidad de software personalizado no tiene límites. Al reducirse el coste de la ingeniería de software gracias a la IA, se creará más y mejor software, no menos. 4️⃣ La experiencia sigue siendo clave: Los mejores ingenieros no son los recién graduados, sino los que tienen una gran experiencia y entienden profundamente la arquitectura de software y los marcos conceptuales. 🚀 La combinación ganadora: Un desarrollador con experiencia que además domina las nuevas herramientas de IA será inmensamente valioso y superior a ambos extremos: el ingeniero experimentado que no se adapta y el recién graduado que solo domina la IA sin tener una base conceptual sólida. Boom!

#STARTUPS 🚀 | M.IA gana la segunda edición del Cofidis Startup Booster junto a otras tres finalistas: Stelo.ai, VirtualBroker y Dost. Estos son sus proyectos disruptores 👇 ✍️ @beazmi 🔗 @Cofidis  f.mtr.cool/mszljmrurv

🔥El 2025, els incendis forestals a la UE ja han cremat quasi 4 vegades més terreny que la mitjana dels 19 anys anteriors. 🌡️La tendència ve acompanyat d'un augment sostingut de la temperatura. Aturar el canvi climàtic és una necessitat. Més: europa.eu/!rWVrwc

@DanielBlancoSWE me acabo de despertar, pesan las décadas

Son las fiestas del pueblo, salí anteayer y hoy aún no estoy del todo recuperado. Señales de que uno se va haciendo viejo 😂👴

@mckaywrigley @EricBuess yep

My honest GPT-5 review: - It is a *phenomenal* everyday chat model - I will default to it for all normal chats - API pricing is incredible, major points here But code? I will still be using Claude Code + Opus.

@Shpigford context7

claude code users: what are the most impactful MCPs you regularly use?

🎓 Catalunya bat rècords en estudis de doctorat: 🔹 18.817 estudiants el curs 2023-2024 🔹 Més del 50 % són dones 🔹 Gairebé el 40 % són internacionals 📣 Els estudis de doctorat es consoliden com una opció de futur entre els estudiants catalans 🔗gen.cat/3J4sbNG

You're absolutely right.