Congratulations on the new model , without any doubt that this model is the top of the art for coding , i have been doing with it a lot of different apps in python with its help , on 2.5 after a long chat with some bugs on the way it got a bit messy and started to fail because the 2.5 pro was looking back to all history in chat , when that happened i had to create a checkpoint and start a new sandbox , paste the checkpoint and continue from there , with this 3.0 today we made a very complex dns safe app that stores the ips from websites or emails recieved or send from my server to a local phonebook in case in future a global dns issue appears so i do not stay in the dark if dns servers fail by some reason to deliver the ip from a domain requested , in normal terms this looks a very simple strategy and code , but adding and openwrt router in the middle of the process where it sends the collected domains to my remote server so it can add them to phonebook and in case a dns issue occours the openwrt calls back from remote server a specific ip for a specific domain it is a complex strategy mostly without opening a port on the router on wan side to do the job . This model is perfect until now , sadly the daily tokens are not enough 
, thank you google AI denvelopers for this excellent model , it is perfect .
Hi @pedropt,
Thank you for sharing your experience with Gemini 3.0! It is great to hear that the model is performing beyond expectations, especially regarding the complex logic required for your OpenWRT and DNS failover project.
Regarding the token limits, since your daily usage is high, I recommend subscribing to the higher rate limits to match your use case. You can review the details here: https://ai.google.dev/gemini-api/docs/rate-limits#current-rate-limits.