[OpenAI Excels at Saving Tokens: GPT-5.6 Sol Uses Half the Tokens to Approach Fable 5] According to monitoring by Beating Insights, Artificial Analysis has released the latest token efficiency comparison. This month, more than five labs have launched new models intensively, yet OpenAI still dominates most of the Pareto frontier. The so-called Pareto frontier refers to the state where no other model can achieve the same level of intelligence with fewer tokens. GPT-5.6 Sol scored a maximum of 59 points, with an average token output consumption of approximately 15,000 per task. Claude Fable 5 scored 60 points, consuming about 33,000 tokens. The output unit price of GPT-5.6 Sol is $30 per million tokens, which is not cheap. However, in this test, the cost per task is still only $1.04, roughly one-third of Fable 5's cost. OpenAI's own Terra and Luna models did not make it into this efficiency frontier. [Original Link]