Ox Alpha is GLM-5-3-Flash

Ox Alpha which was viral for last couple of days (which has provided almost unlimited inference in stealth mode) is actually GLM-5.3-Flash

Some of the key highlights:

• Cheaper cost with better capabilities. Input: $0.15, Output: $0.50, Cached input: $0.03 (Standard API pricing)
• Has Vision/Image support with 1M context window
• MIT License
• It ran entirely on Chinese AI chips when it was in preview mode as Ox Alpha.
• It is as good as Opus 4.8

Its weights are available in HuggingFace

Reference

• https://x.com/Zai_org/status/2092616204787626030
• https://artificialanalysis.ai/models/glm-5-3-flash