I strongly echo the community’s concerns about the removal of the detailed Gemini thinking process in AI Studio. The raw CoT were critical for understanding model reasoning. The summarized output lacks the depth needed for effective use. Please consider reinstating the 03-25 model(it was also significantly stronger than 05-06 model in many fields, the quality regression is palpable) and raw CoT to restore these valuable tools for developers and users.
2 Likes
Yes the IQ degradation of 05/06 is significant from 130 of 03-25 to 109 of 05/06, even if coding improved it make NO Sense to help complex coding task like minecraft modding or Mcreator development. I hope newer model can fix core function shut down, Not just creative writing but reasoning ability
2 Likes
The IQ degradation of 05/06 is significant from 130 of 03-25 to 109 of 05/06, even if coding improved it make NO Sense to help complex coding task like minecraft modding or Mcreator development which invovle in complex thematic brainstorming like what creative writing does . I hope newer model can fix core function shut down, Not just creative writing but reasoning ability
3 Likes
That is a very terrible model as I must admit, I am totally flamed by its performance days ago which just like a kid breaking entire story logic. Thankfully most of job was down by 03/25 and 03/25 exp. See this Gemini 05/06 children-level understanding for Hard-sifi setting. I sincerely hope their next update are not going to do same regression over and over again
1 Like