What do you think about the latest 3.7 flash? It seems to significantly cross Sonnet 5 in benchmarks, does it seem decent during usage?
What about the quota, as it provides ~10x lower cost until the end of the year. Do you feel that the usage consumes less quota?
I would really appreciate any feedback
When Gemini 3.5 Flash first came out, I tried using it, but it wasn’t productive at all—it was just too buggy. I stopped using Antigravity and moved over to Chatgpt 5.6 luna and sol, then used Open Code for a while, and more recently Codex.
​However, this latest version 3,7 genuinely impressed me. It might not be a top-tier frontier model, but its coding capabilities have improved drastically. I’ve gone back to using it heavily because it’s working so well now: it catches bugs accurately and handles planning properly. It might not be the absolute number one, but it’s leagues ahead of where it was—and, above all, it is blisteringly fast."
Thank you! I’m busy on some other work so I can’t try it myself, but if it works good like Luna or Terra, that’s pretty excellent
Blisteringly fast", my ass! 
30 seconds of pure thinking. Classic Google backend. 
Let me test it out through Agy and see what happens
