Hello everyone,
I’m currently facing a frustrating issue with date calculations in GPT-4.1 (and other models too). As part of our conversation logic, the model is expected to calculate the number of business days between two dates. However, it consistently produces inaccurate results.
For example:
- When asked to calculate business days between July 14 and August 18, it returns 29 instead of the correct 26.
- Similarly, when the model is asked to determine the end date after 37 business days starting from August 11, it incorrectly outputs October 10 instead of the correct September 30.
Despite trying multiple prompt variations and formats, the model’s behavior remains the same.
Has anyone else experienced this or found a reliable solution/workaround?