User satisfaction reportedly divides AI labs more than task success

Agent Arena says it tracked the best agents from six AI labs from late June to early October. On praise versus complaints, OpenAI and Anthropic were 35 points above its leaderboard average; Moonshot, xAI and DeepSeek were 5 to 13 points above. It says the average lab's best model went from -4 to +14 points on confirmed success, while gaps on Bash recovery and steerability narrowed.