OpenAI’s GPT-6.1 Terra (Max) lands at #8 in Agent Arena with a 10.6% net improvement

OpenAI's mid-tier model posts a median task cost of $0.62, making a case that cheaper can still be competitive OpenAI’s GPT-6.1 Terra, running at its Max setting, has placed #8 in Agent Arena. It posted a 10.6% net improvement and a median task cost of $0.62. What the numbers show Agent Arena ranks AI models on agentic work. These are tasks where the model acts on a user’s behalf rather than simply answering questions. Terra (Max) came in eighth overall. The 10.6% net improvement is the headline metric from the ranking. Next to it sits the cost figure: a median of $0.62 per task. Broader evaluations put Terra’s median task costs between $0.40 and $0.62. The Agent Arena result sits at the top of that range. That fits a Max configuration, which trades extra compute for better output. The model has posted solid scores elsewhere as well. On Terminal-Bench 4.0 it scored 50.3%, compared with 21.5% for its predecessor, GPT-5.6 Terra, at maximum effort. Terra (Max) also achieved a 60% pass rate on the Agents’ Last Exam at max effort. It scored around 74% on OSWorld, a benchmark that measures how well a model can operate a computer the way a person would. Where Terra fits in the GPT-6 famil...