Arena says OpenAI’s GPT-6 Astra is available to test
Arena invited users to try Battle Mode and Agent Mode, saying their votes would shape the evaluation. Its September 4 announcement said scores were coming soon.
TLDR
Arena announced on September 4 that OpenAI’s GPT-6 Astra had joined its platform, with testing in Battle Mode and Agent Mode. Arena says its agent evaluation measures models on millions of real-world tasks, allowing them to use web search, filesystem and terminal tools to complete complex workflows. User votes would shape Astra’s evaluation, Arena said, with scores still to come.
Combined views
3.7K
2 Sources, first seen 26d ago