Tibo reminds AI not only token: GPT-5.6 SOL is more expensive, but the bill is lower

Tibo cautions that when comparing prices for different AI models, one cannot look at "how much per million token". 766 tokens were used on GPT-5.6 SOL, while Claude Opus 5 estimated that 1170 were needed, approximately 34.5 per cent less. The current standard API input price for both models is US$ 30 per million token 5 output, US$ 25 per million. Opus 5 is cheaper on a single export unit price, but SOL spends about $0.023 and Opus 5 around $0.029 at an output rate of about 21 per cent. Tibo uses pizza to compare it more rationally with "how much does it cost to complete a mission" and to take into account the success of the model and actual token。