Ox Alpha on OpenRouter identified as GLM-5.3 variant with alignment tax
airesearch12 · x · 2026-08-21
Analysis suggests the stealth model "Ox Alpha" on OpenRouter is a variant of Zhipu's GLM-5.3. While trait differences are minimal, Ox Alpha shows increased alignment and safety features—such as a more clinical register and sanitized output—indicating additional post-training. This "alignment tax" results in a significantly lower ELO score on EQ Bench 4 compared to GLM-5.3, likely a trade-off to mitigate risks for open-weights release. The model remains outstanding despite the drop.
More from Models
- Ox Alpha Coming Soon, Excitement Builds — zephyr_z9 · 2026-08-21
- Ox-Alpha's take on researcher Janus: real contributions, polarizing figure — jd_pressman · 2026-08-21
- mLateOn matches 8B models on Japanese JMTEB, beats Japan-specific models — IgorCarron · 2026-08-21
- User Report: Opus 5 Regresses, Context Compaction Losing Track — springrod · 2026-08-21
- The Forecasting Company open-sources t0-alpha, detailing the 5-era evolution of forecasting models — fpedregosa · 2026-08-21
- SentenceTransformers V6 Unifies Text, Vision, and Audio in Single API — ManuelFaysse · 2026-08-21