Hi Fatika! I read your EA forum post and the project is really interesting! May I ask, why are you choosing API calls for older closed source AI models (E.g. Claude sonnet 4) rather than the newest models such as Claude sonnet 4.6 and Opus 4.7, at a budget of $500 you should be able to call those instead which would make this benchmark a lot more valuable. I'm guessing that current African agricultural systems are using free open-source models but knowing the capabilities of frontier AIs will give you a good idea of what open source models will be capable of within a year or two at most.