AI brief
An evaluation of the CoT-Control suite on GLM, Kimi and DeepSeek models found that Kimi and DeepSeek models have poor chain-of-thought control, while GLM models were outliers in their ability to control it, with further investigation into GLM-5.1.
Written by AI from LessWrong's published text.
Background
From our own funding, model and news trackers, not from the source.
DeepSeek
- Latest funding
- $15B · 6 Oct 2026
- Newest model on the API
- DeepSeek V4.1 Flash 1017 · 17 Oct 2026 · $0.15 in / $0.6 out per million tokens
- Last 30 days
- 19 news stories
Source
“TL;DR. We run the CoT-Control evaluation suite on GLM, Kimi and DeepSeek models and find that while Kimi and DeepSeek models have poor CoT control, GLM models are outliers in their ability to control CoT.”
Interpretation or community commentary rather than straight reporting.
Read original story ↗