Retweet of GPT-6-Astra Multi-Agent Coding Evaluation Claim

OpenAI research engineer Ted Sanders retweeted a statement from leo_linsky. According to the post, an evaluation of GPT-6-Astra took place across 100 complex unsaturated multi-agent coding environments. In these settings the model both competed and cooperated with other participants. The announcement focuses on the use of intricate scenarios that involve multiple agents interacting dynamically. Visible replies and the conversation around the post do not include additional confirmation or results from the testing. The claim originates directly from the original poster as shared in the retweet.