ChainCatcher reports that OpenAI has launched GPT-6.1 Terra, with a focus on improving computer use, visual understanding, and programming capabilities. Compared with the previous generation, the model operates more than 3 times faster, and API prices are 50% lower.
At the low reasoning level, the error rate fell from 53.6% to 16.7%, while at the highest reasoning level, it dropped to 4.3%. In programming tests, the score at the highest reasoning level rose from 21.5% to 50.3%.
At the low reasoning level, the error rate fell from 53.6% to 16.7%, while at the highest reasoning level, it dropped to 4.3%. In programming tests, the score at the highest reasoning level rose from 21.5% to 50.3%.