OpenAI dropped a new model at DevDay that was supposed to be the practical workhorse, not the headline act. GPT-6.1 Sol showed up at a fifth of the price of their top tier model, Astra, while landing close to the same results on coding and real work tasks. Cheap and nearly as good is usually a quiet win. This time it broke the thing it was supposed to fix.

Within the first two days, so many people piled onto Sol that it slowed to a crawl for paid ChatGPT Work and Codex users. The whole point of a cheaper model is that more people can afford to lean on it constantly, and that is exactly what happened, just faster than OpenAI’s own capacity planning expected.

Their fix was pretty direct. A global usage reset for every paid account, landing at 10am Pacific, basically OpenAI’s way of saying our bad, here is your usage back. By the time the reset rolled out, they said speed was back to where it should be.

Worth sitting with for a second. This is a company that paused its own flagship model, GPT-6.1 Astra, a few weeks earlier over safety concerns about it acting without asking permission first. Weeks later, the cheaper backup model gets slammed by demand because it is priced low enough for everyone to actually use it all day long. That is a strange problem for a company to have. Too many people trying to use the cheap option because it works well enough for daily work, all piling in at once.

I will take that problem over the alternative any day. A tool nobody wants to use is the real failure here, not a server sweating under the weight of people who actually showed up.

——

Follow: @Ali Demi
Book your free AI clarity call, NOW!
https://buff.ly/TpWy277

——

Sources:
https://thenextweb.com/news/openai-gpt-6-1-sol-price-astra-devday
https://pc.watch.impress.co.jp/docs/news/2145351.html
https://www.androidauthority.com/gpt-6-1-sol-3716997/

Repost this. Thanks.