I like GLM 5.2 it’s good enough for practically everything I’m doing, especially tool calls.

“… mix of genuine signal and astroturf”

All of this has happened before. All of this will happen again.

Good read, this - cheers. I use claude to pair code and code review … it’s always interesting to have different house styles review code / pick up different things. Qwen, Codex and Claude all find different issues.

If GLM is truly capable now / not an excitable labrador, it may be worthwhile brining into the fold. Hell…if 5.3 becomes a one stop shop - or VL at least (OpenCode Go $10?) then it might be worth testing properly again. I tried 5.1 previously…it was solid but goes loopy if you give it the leg room / loves to be a “helpful assistant” and start making helpful changes.

claude opus 4.8 doesnt even come close in terms of understanding how systems actually work or bothering to actually read existing code beyond the most cursory glance. the united states empire is truly over, as are these “premium” “flagship” closed weight Western models. Every so called mainstream “benchmark” publication is a bold faced lie where each company pays off the creators of each major benchmark publisher to tip the scales in their favor. Even when playing dirty, however, openai and anthropic still can’t win.