Phil

Pls tell me we at least get a clear understanding how much percentage we will have used compared to the reset time. It's already a bad user experience with the 5h limit not telling you how much is left. For a 7 day limit, this impossible to plan when to use Opus and Sonnet…

Not only is the Max 20x no longer 4 times as much usage as the Max 5x plan, it's not even twice as much. Is having 2 accounts with a $100 Max 5x plan better value now? (I only care about opus)

why does Claude constantly think about if the code is malicious? What did they do to its system prompt?

People are still underestimating how much better Claude Code is compared to anything else on the market. Huge repo you know nothing about? No problem Claude will search up all interfaces, patterns used, helper functions, you don't need to understand the structure at all. Library…

What are you using ChatGPT agent for? Haven't found any good use cases yet, might be a prompting skill issue tho

Chatgpt Agent is sadly just as bad at generating slides as most other models. Here are a few slides from a RAG survey presentation it generated. Completely unusable, looks like the first presentation made by a computer illiterate boomer, just plane text not aligned, zero styling,…

Claude Code native Windows support! Huge day for .NET developers

Kimi K2, specifically for agentic coding?! This might be another R1 moment!

So Grok 4 just feels similar to o3? What's supposed to be the selling point here? Did not watch the livestream, just tested it, am I missing what makes it special?

"I don't see much difference between Opus & Sonnet." True: single edits feel similar. But agentic coding isn't about single edits. It's about compounding accuracy. A "small" 3pp benchmark improvement means the difference between code that meets requirements, and hours spent…

I don't even want a Claude Opus 4.1, all I want is Opus with a 500k context window

what? guess SQL for production orders is too dangerous for the constitutional classifier? please don't shut down my claude account tho

wtf it also works with image generation!? the canvas is calling the imagen-3.0 api here, all inside the canvas!

Wait you can now integrate the Gemini api directly into the canvas by pressing this button?! gemini changed my example to generate new quotes using 2.0 Flash directly in the canvas using an api key provided by the gemini app! This opens so many new cool ideas!

If the voice mode does not include this setting, the chances that I will ever use it drop by 90%. Slow-talking voice modes are the most annoying part of AI apps. I'm looking at you, Google and Perplexity

Guide to build an AI slide generator that actually works: 1. Claude Opus 4 - prompt it to generate static html + css, no javascript. each slide is its own div with a specific id. each slide div has exact css dimensions 2. script that does static analysis to check if any text…

Can we please just get a setting for thinking token in Claude code? The think, think harder, ULTRATHINK is a fun idea but it gets old pretty fast when you have to type ULTRATHINK after each message to get the best performance. In fact, opus started ending its messages with…

Either ChatGPT has gotten way worse with the "it's not (only) x–it's (also) y" writing pattern in the past few months, or many more ppl are using unedited ChatGPT output for serious posts and articles. This is honestly turning me into an AI boomer, come on, at least edit the em…