Tag
Ox Alpha, a free model with 1M context, achieved a ~63% score on the DeepSWE benchmark, matching frontier mid-tier models and suggesting potential for local AI agent development.
The article reveals that Ox Alpha is based on z.ai's unreleased GLM model, verified through matching token sizes in the tokenizer and image encoder.
Nous Research announces that Ox Alpha is free via Nous Portal for a limited time, with the portal offering access to numerous AI models at varying token prices.
A Twitter user shares a positive experience using Sol to identify Ox-Alpha, highlighting its swift and accurate analysis.
The article speculates about the Ox Alpha stealth model, comparing it to GLM5 Air and Mimo V3.