TeortaxesTex Compares DeepSeek V4 Pro and Ox Alpha Benchmarks
Posts from teortaxesTex compare Ox Alpha performance to DeepSeek and speculate on its GLM model lineage.
TLDR
TeortaxesTex posted that DeepSeek V4 Pro leads a CVE rediscovery benchmark ahead of Ox Alpha. The same account suggested Ox Alpha could be a post-trained descendant of GLM 5V Turbo, though details on its core model remain unknown. Rohit Krishnan replied that Ox Alpha performs much better than Gemini 3.7 Flash. The conversation centers on these model comparisons and possible connections among GLM, DeepSeek, and Ox Alpha based on the visible posts.
Combined views
45K
6 Sources, first seen ago
Sources
T2 
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTexThat's quite good for ox that said, DeepSeek does amazingly here https://twitter.com/pilvar222/status/2091155448590160048
- likes: 114
- replies: 25
- bookmarks: 19
- reposts: 1
T2 Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex> This model appears to be a "descendant" of GLM 5V Turbo …ox can literally be a post-trained 5V Turbo. We don't know anything about its core language model though. https://twitter.com/Lentils80/status/2092001912170156067
- likes: 104
- replies: 3
- bookmarks: 17
- reposts: 3
BR BR Bindu Reddy@bindureddyOxAlpha is GLM 5.3 Flash It seems like Nvidia may have provided the compute to give away 100T tokens for free!! Genius marketing of open-source AI!
- likes: 425
- replies: 26
- bookmarks: 27
- reposts: 10