@antirez There is a small office with 20-30 people and good wifi network. Each person has a MacBook, but not a beefy one, just 32-64 GB. Theoretically, can they unite to do distributed chat completions at good enough speed?
I just gave @claudeai and HTML one more chance
And guess what it produced?
- unreadable white text on beige background
- schematic view is still in plaintext
Now I'm writing this tweet, rewinding or fixing the CSS instead of getting the promised rich HTML and staying on my task
"Explain schematically data flow and logic" works so much better than HTML with @claude Opus:
- plain text, less tokens
- no need to debug/fix html/css (like white text on white bg pretty often)
- less cognitive load than HTML
Works with Opus only. GPT-5.5 sucks at it.
"Explain schematically data flow and logic" works so much better than HTML with @claude Opus:
- plain text, less tokens
- no need to debug/fix html/css (like white text on white bg pretty often)
- less cognitive load than HTML
Works with Opus only. GPT-5.5 sucks at it.