It's about $1 per thousand tweets and access to 0.3% of the total volume. I think the subscription is 50M "new" tweets each month? There are other providers who continually scrape Twitter and sell their back catalogue.
Researchers are complaining that it's far too high for academic grants. Probably true, but that's no different from other obscenely priced subscriptions like access to satellite imagery (can easily be $1k for a single image which you have no right to distribute). I'm less convinced that it's impossible for them to do research with 50 million tweets a month, or with what data there is available. Most researchers can't afford any of the AI SAAS company subscriptions anyway. Data labelling platforms - without the workers - can cost 10-20k a year. I spoke to one company that wouldn't get out of bed for a contract less than 100k. Most offer a free tier a la Matlab in the hope that students will spin out companies and then sign up. I don't have an opinion on what archival tweets should cost, but I do think it's an opportunity to explore more efficient analyses.
https://www.wired.com/story/twitter-data-api-prices-out-near...
Researchers are complaining that it's far too high for academic grants. Probably true, but that's no different from other obscenely priced subscriptions like access to satellite imagery (can easily be $1k for a single image which you have no right to distribute). I'm less convinced that it's impossible for them to do research with 50 million tweets a month, or with what data there is available. Most researchers can't afford any of the AI SAAS company subscriptions anyway. Data labelling platforms - without the workers - can cost 10-20k a year. I spoke to one company that wouldn't get out of bed for a contract less than 100k. Most offer a free tier a la Matlab in the hope that students will spin out companies and then sign up. I don't have an opinion on what archival tweets should cost, but I do think it's an opportunity to explore more efficient analyses.