4/13/2026 at 6:24:36 PM
by mikenew
4/14/2026 at 7:21:33 AM
Anybody running GLM 5.1 locally or within their company? What kind of hardware do you need to achieve performance similar to Claude?by cedws
4/14/2026 at 12:59:50 AM
Unjustified from the perspective of a consumer. Coding api is incredibly slow, frequently times out, and model quality is nowhere near Claude Opus in actual usage. Usage quotas also fill much sooner than expected given the claimed ratios to Claude plans. No value compared to Codex or Claude. Quite a surprising change.by ungruntled
4/14/2026 at 3:28:30 AM
I've found GLM 5.1 to be extremely good, as long as you keep the context under 100k or so. But I do agree; the actual service is rough. It's often unusable during peak hours.I imagine they're trying to slow down the acquisition of new customers because they're so overloaded. But yeah at double the price it doesn't seem worth it anymore.
by mikenew
4/13/2026 at 6:24:36 PM
I can't find an announcement on this, but here's the archived subscription page from a couple days ago: https://web.archive.org/web/20260410092340/https://z.ai/subs...The Max plan has doubled, and the Lite and Pro plans have more than doubled. No change in usage limits.
by mikenew
4/13/2026 at 6:38:32 PM
itsby dotcoma