The AI model Ox Alpha processed 11.6 trillion tokens in 72 hours on the OpenRouter platform, approximately 2.6 times the previous record for a model launch. This massive volume is attributed to automated coding agents that ingest entire codebases and perform repeated cycles of reasoning and generation. The model, launched anonymously on August 20 with a one-million-token context window, remained heavily used even after the free preview period ended on August 24. Early benchmarks show success rates above 80% on coding tasks, outperforming some established competitors. The provider’s identity remains unconfirmed, though speculation points to Z.ai’s GLM model family.
Source: Read the original article
