@zachary-1.bsky.social
Corporate culture seems to have rubbed off on some of these AI models. The amount of sycophancy is unparalleled.
February 10, 2026 at 11:30 PM
Corporate culture seems to have rubbed off on some of these AI models. The amount of sycophancy is unparalleled.
Okay Claude Opus 4.6 is awesome, though expensive!
It seems to understand nuances very well.
In the last few days I've tried a few, so far Opus 4.6 > Kimi2.5 > GLM4.7. If logic is critical and risk is high then go with Opus, you'll sleep better at night.
It seems to understand nuances very well.
In the last few days I've tried a few, so far Opus 4.6 > Kimi2.5 > GLM4.7. If logic is critical and risk is high then go with Opus, you'll sleep better at night.
February 10, 2026 at 9:05 PM
Okay Claude Opus 4.6 is awesome, though expensive!
It seems to understand nuances very well.
In the last few days I've tried a few, so far Opus 4.6 > Kimi2.5 > GLM4.7. If logic is critical and risk is high then go with Opus, you'll sleep better at night.
It seems to understand nuances very well.
In the last few days I've tried a few, so far Opus 4.6 > Kimi2.5 > GLM4.7. If logic is critical and risk is high then go with Opus, you'll sleep better at night.
New consideration, keep script logic minimal.
More lines of code means greater chance the model will flub on modifications.
More lines of code means greater chance the model will flub on modifications.
February 10, 2026 at 8:10 PM
New consideration, keep script logic minimal.
More lines of code means greater chance the model will flub on modifications.
More lines of code means greater chance the model will flub on modifications.
Kimi2.5 > Claude 4.5 models. How's 4.6 performing?
February 8, 2026 at 8:45 PM
Kimi2.5 > Claude 4.5 models. How's 4.6 performing?
I had a phenomenal experience with kimi2.5, up until today, when it's full power was channeled into creating a disaster.
February 4, 2026 at 1:25 AM
I had a phenomenal experience with kimi2.5, up until today, when it's full power was channeled into creating a disaster.
Okay kimi 2.5 is incredible, currently my default model going forward. Atleast for coding.
February 1, 2026 at 4:15 AM
Okay kimi 2.5 is incredible, currently my default model going forward. Atleast for coding.
Between OpenCode and Kimi2.5, I've decided it's feasible to start winding down claude code usage.
It's an incredible tool but the cost feels increasingly untenable.
Anyone else exploring or have adopted alternatives?
It's an incredible tool but the cost feels increasingly untenable.
Anyone else exploring or have adopted alternatives?
January 31, 2026 at 10:55 PM
Between OpenCode and Kimi2.5, I've decided it's feasible to start winding down claude code usage.
It's an incredible tool but the cost feels increasingly untenable.
Anyone else exploring or have adopted alternatives?
It's an incredible tool but the cost feels increasingly untenable.
Anyone else exploring or have adopted alternatives?
Okay, it's settled I'm done with drinking until $1M ARR.
Sometimes you got to be ruthless about your habits to get what you really desire.
Sometimes you got to be ruthless about your habits to get what you really desire.
December 22, 2025 at 5:15 PM
Okay, it's settled I'm done with drinking until $1M ARR.
Sometimes you got to be ruthless about your habits to get what you really desire.
Sometimes you got to be ruthless about your habits to get what you really desire.
Anyone know of a github-like platform for mechanical engineering contributions?
December 16, 2025 at 5:24 PM
Anyone know of a github-like platform for mechanical engineering contributions?