Introducing EmbeddingGemma 2! 🚀 Our lightweight, multimodal embedding model maps text, code, images, video, and audio into a single, unified embedding space. Optimized for on-device use cases, it features: - 740M parameter form factor with modular encoders - Flexible dimension sizes (768dim-128dim) via Matryoshka Representation Learning (MRL) - 8K context window (4x larger than text-only EmbeddingGemma) - A commercially permissive Apache 2.0 license

Software architecture is easy... Group related concepts together = build better applications Can it be that simple? It's not, of course. But it is directionally correct. Another name for this is cohesion. We want to maximize how close together files for a single feature are. Layered architecture typically has low cohesion. Why is this a problem? It forces you to make changes in multiple different layers to implement or update a feature. Vertical slices take a different approach. All the files for a single use case are close together, which increases their cohesion. It's easy to find all the relevant components for a feature. Thus, making changes is simpler. You can learn more about VSA here: milanjovanovic.tech/blog/ver…

🚀 New in Symfony 8.2: Performance Improvements ➡️ symfony.com/blog/new-in-symf…

❌ En serio, NUNCA más compres fondos indexados en estas plataformas: · BBVA (150€/año por cada 10.000€) · CaixaBank (150€/año) · Bankinter (113€/año) · ING (110€/año) · Santander (entre 74€ y 94€/año) ✅ Plataformas que SÍ deberías usar ↓↓

Today we’re introducing Gemini 4 Argon. It delivers frontier performance in complex workflows across real-world software engineering, knowledge work, and cybersecurity defense with an industry-leading 1M token output limit.

Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. Here’s a look at the benchmarks:

NVIDIA drop a 3B vision-language model for fast, high-quality visual grounding in real time accurately. - Parallel box decoding - 10× faster than Qwen3-VL - Trained on 138M queries/785M boxes - GUI, OCR, and document layout, dense detection - Open source Useful for computer-use agents and Physical AI,

Introducing Claude Sonnet 5.5, the second model in the Claude 5.5 family. It’s a clear upgrade over Sonnet 5, runs more than 30% faster, and costs up to 30% less for most work.

AGENT WARS We asked 5 top AI models to design, engineer and 3D print a bridge to with the goal of holding the most possible weight. 🇺🇸 Anthropic: Claude Opus 5.5 (High) 🇨🇳 Kimi: Kimi K3 (High) 🇺🇸 Meta: Muse Spark 1.3 (High) 🇺🇸 OpenAI: GPT 6 Astra (High) 🇺🇸 SpaceXAI: Grok 4.7 (High) Each bridge had specific parameters. It had to span 2 feet, use a maximum 500 grams of filament, and it had to be under 18 hours of printing time. We specified what type of weight and where we would place it on the bridge. The Score: • Claude Opus 5.5 held an estimated 130 lbs! It took 9 hours and 11 minutes to print 17 parts and used 441 grams of filament. • Meta Muse Spark 1.3 held 26.5 lbs. It took 13 hours and 12 minutes to print 49 parts and used 478 grams of filament. • OpenAI GPT-6 Astra held an estimated 17.5 lbs. It took 15 hours and 44 minutes to print 29 parts and used 442 grams of filament. • SpaceXAI Grok 4.7 did not finish. On assembly it could not stand up by itself. It took 12 hours and 22 minutes to print 29 parts and used 460 grams of filament. • Kimi K3 did not finish. It wasn’t engineered correctly so the bridge could not even be assembled. It took 11 hours and 19 minutes to print 35 parts and used 446 grams of filament. That means our champion for this episode is Claude Opus 5.5. Not only did it hold the most weight (by FAR) it also took the least time to print, used the least amount of parts and used the least amount of filament. This was episode 02 of Agent Wars, stay tuned for episode 03.

If you're using AI agents for coding, Architecture Testing is a must-know. Here's why. 👇 Architecture testing helps you control technical debt. Technical debt is a consequence of prioritizing development speed over well-designed code. And you can bet AI agents churn out code faster than you or I ever could. With architecture testing, you can control: - Direction of dependencies - Naming conventions - Various design rules I wrote this in-depth guide that I think you'll enjoy: milanjovanovic.tech/blog/enf… You can even tell your agent what your architecture rules are, and it can write the tests to enforce them. Then you just run this as part of your CI, and you'll catch any issues much faster. Would you add this to your code review process?

Today at @WeAreDevs, Docker and the @linuxfoundation are announcing a collaboration around the Docker Sandbox Kit Specification: an open standard for declaring what an agent may do, where it may reach, and what it may touch. Open source under Apache 2.0. Read the deep dive: bit.ly/4rrcjql

Upgrading a global app from Symfony 4 to 7 doesn't need a huge team 💻 See how Yuka automated AST refactoring, and modernized their high-traffic API while giving back to open source! ✨ symfony.com/blog/case-study-…

Run open models like Gemma 4 completely offline in the Antigravity SDK. Built on Google AI Edge’s LiteRT, you can now run Gemma 4 directly on your local GPU. Zero API costs, total data privacy, and no internet required.

¡Ya sacaron una alternativa de Jev de código abierto! Se llama Kev y tienen tres tamaños diferentes. ✓ Se ejecuta en tu ordenador. Los datos no salen de ahí ✓ Responde sí o no, elige una opción o pone una nota ✓ Perfecto para clasificar → github.com/jaredpalmer/kev

It's 2026, and you're still not using OpenTelemetry? Here's how this makes your life as a software engineer easier. If you care about observability, you should know about OpenTelemetry. It's a vendor-neutral, open-source standard for instrumenting applications to generate telemetry data. That's a few buzzwords to say: logs, traces, and metrics. Does it work in .NET? Yes, it's supported with a few libraries that are easy to integrate. I wrote a simple guide to getting started with OpenTelemetry. Learn more here: milanjovanovic.tech/blog/int… What do you use to implement observability in your apps? Once you have OpenTelemetry set up, the next step is to collect and view the data. For this, I usually use the Grafana stack with Loki, Tempo, and Prometheus.

Just merged support for Jev into symfony/ai - from now on that mono repo gets split into exactly 100 packages. We added way more: support for Higgsfield, Together AI, Fireworks, Venice, and next up Eden AI. And support for async jobs/batch and agents using tools of MCP servers.

I have upgraded my SF Observability Demo Application If you are lost with Grafana, Loki, Elasticsearch, Kibana, Clickouse, or Redash, this demo is made for you 🔥 You'll get a fully provisioned o11y platform in one command, with beautiful dashboard 📊 github.com/lyrixx/symfony-ob…

Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.

This is insane try.cloudflare.com

Hatte vor ein paar Tagen die Ehre den Betriebshof der Busse von Oviedo in Asturien (Spanien) besuchen zu duerfen. Mein Respekt. 70 Elektrobusse. Nur noch 9 Dieselbusse. Diese Dieselbusse fahren auch nur deshalb noch da es fuer bestimmte Strecken keine geeigneten Modelle mit E-Antrieb gibt (man braucht fuer bestimmte Strecken kleine sehr schmale Busse die reichweitenstark sind). Lieferant der E-Busse ist Mercedes (E-Citaro). Ladestationen von X-Charge. Der Betriebshof ist an die Mittelspannung angeschlossen. Es sind einige Transformatoren im Einsatz die die 400 KW CCS Ladestationen powern. Das Programm war von langer Hand sorgsam geplant. Nochmal Glueckwunsch zu diesem sehr schoenem Erfolg. alsa.es/-/tua-prueba-el-prim…