GLM 5.2 Preset Miscalculates by 2.667x
LiorOnAI · x · 2026-07-12
A hands-on test tasked four frontier models with building a KV-cache debugger, demanding exact formulas without cutting corners.
The results revealed:
- All four models nailed the complex arithmetic parts.
- However, a GLM 5.2 preset was off by 2.667x due to an incorrect layer count setting, failing silently without any warning.
- The other four presets functioned normally.
The original poster highlighted the irony that the cheapest model was the sole one to make a mathematical error.
More from Models
- giffmana: the env being used in training is part of the point — giffmana · 2026-09-11
- awesome-llm-leaderboards: an open-source directory of LLM leaderboards, pricing tables, comparison tools — Last_Establishment_1 · 2026-09-11
- Anthropic claims it works to keep eval environments unidentifiable to models — MaxKannen · 2026-09-11
- Nex N2.5 Pro released on Hugging Face with 407GB of weights — jinnyjuice · 2026-09-11
- RoMa v2 image matching model unveiled in the usual black poster — ducha_aiki · 2026-09-11
- OpenAI rated Astra 'Critical' for cyber capabilities — and admits it's harder to monitor — theguywhobuilds · 2026-09-11