It’s crazy to me that you can generate 100 pictures for just $3

Interesting to see that Prisma got faster after dropping Rust because queries no longer required serialization between TypeScript and a separate Rust binary. With the new TypeScript + WASM Query Compiler you have: >everything in one runtime >no serialization overhead >smaller

I bookmark things all week, and the weekend is for actually reading them and clearing my list. Feels so good when its finally empty gain

Render a photorealistic animal figurine, sitting on a shelf with tiny accessories like a food bowl or toys, in miniature sizing and bright colors.

In the style of Anime, depict a peaceful city street with soft light, quiet shops, and a few people strolling leisurely. The scene should feel calm and inviting, with gentle colors and a serene atmosphere that highlights the slower side of urban life.

A Pekingese dressed in elegant Hanfu, riding a skateboard through a bamboo forest, painted in the style of traditional Chinese ink wash painting (shui-mo hua).

A cozy Hamburg coffee shop on a rainy day, gray Elbe light through big windows, cobblestone reflections outside, bicycles parked, people on laptops, barista steaming milk, Nordic-minimalist with HafenCity industrial touches.

A hyper-realistic portrait of an elderly woman with deep wrinkles, silver hair in a bun, wearing a red scarf, lit by soft golden sunset light, ultra-detailed skin texture

I created a little app today to compare image generation from gemini-2.5-flash and seedream-4.0. Seedream's model compares quite well and so far, for image generation, I like the results more. I created a few images in the comments (left: Gemini, right: Seedream)

Just learned you can create mobile apps with php using @nativephp. Sounds like something I'll have to try this weekend

Today, I used @aisdk for the first time to generate images with gemini-2.5-flash-image. The Google Cloud Console was difficult to navigate, especially since a billing account is required even for a free plan, but at the end I managed to succesfully generate images using the

Just a few reasons to hate Oracle: Predatory audits, $115M data privacy suit, $23M bribery fines, gender pay bias, the Google Java API war and dozens of other controversies

Can we get a larger text field for custom modes in @cursor_ai. Maybe as a modal? It's annoying to write long instructions. I need to switch to an editor to write them and then switch back to paste them

So they actually degraded model quality. No wonder people are switching to other CLI tools

Had to use an internal library today that was not documented at all, I hate when you have to figure it all out by yourself. But thanks to Cursor I can just let him figure it out for me and explain to me how I can use it

I actually enjoy Mondays, I can't wait to dive back into coding at work. It's like starting the week with a fresh mind and full with energy, making it one of my most productive days

I really like that you now get notifications when Cursor completes your request or needs permission to run a command, allowing you to switch tasks while Cursor works on your request

It’s wild how the sentiment around Claude Code flipped, people used to call it the best tool out there, and now my feed is full of disappointment. Did Anthropic actually make changes that ended up making it worse? Not using it rn, I'm still happy with Cursor

As a junior dev, I actually enjoy coming to the office daily. It's easier to communicate with other devs and get help from seniors. I don't understand why new grads would prefer working from home. You're just starting out and so it's much easier to learn and get help in person

Created my own little MCP server today and connected it to Cursor. Learned a lot about the Protocol and basically MCP defines three core primitives for servers: -Tools: Executable functions for AI actions (e.g., file ops, API calls, DB queries). -Resources: Data sources for AI