OAI and Ant cannot win this way
retained text
I currently have both OAI 100 and Ant 20 dollar subs. In recent days, both Astra and today O 5.5 quality have gone down again. Astra introduced mistakes into a project when I made the mistake of switching the session from the reliable DS 4.1 workhorse for just one request. What did it do? It destroyed formatting in an unrelated part of an entry form that 4.1 never had problems with. Of course. O 5.5 has been working on a complex 3d model + animations for me since day of release. That model has been developed by Astra before for another 1-2 weeks. It's a complex female chatbot avatar with speaking and gesture animations and complex hair. It should look realistic. Since yesterday, O 5.5 is running in circles. Takes forever. And the last iteration after one day of work was a step back. The avatar looks like a comic book female, after almost achieving realistic looks before already. Of course. I can't with this anymore. I have given both players more than 5 chances. And they do this. Every time. I would be better off never touching their models again. Also, I have been using Deepseek v4.1 since it came out. Never had problems. It \_just\_ works. Why can't Ant and OAI be this way? Can't they see that they make it impossible to stay with them if they keep acting this unprofessionally?
document fetched 2026-09-29 00:41 · canonical https://reddit.com/r/ClaudeCode/comments/1wsfyjn/oai_and_ant_cannot_win_this_way