I am so confused when I see posts on my timeline about GLM 5.2 beating all the latest frontier models.
Not denying that it isn’t a great model, and awesome that it’s open, but it benchmarks closer to GPT 5.4, and quite far behind current gen models.
Things people get wrong with my grill-* skills:
- Being too passive
- Not grilling in parallel
- Not prototyping
- Going into the dumb zone
- Grilling too hard
- Grilling too large a topic
- Using too dumb a model
- Clearing the context too soon
Here's the breakdown:
The reason agents are so good at Linux is that all 40 million lines of kernel code was part of the pre training. Along with every other open source dependency. This really does make every obscure error message shallow, and the system completely malleable.