Tester says Sonnet 5.5 output looks more realistic than Sol 6
The author compared Sonnet 5.5 and Sol 6 with the same prompt and reference at high effort; both used 1% of the weekly quota and took about 20 minutes, but the author found Sonnet's output far more realistic and questioned how Opus overtook Astra.
What a marketing move from OpenAI: nerfing usage on their $200 tier WHILE Opus 5.5 and Sonnet 5.5 are beating GPT-6 Astra and Sol. 10/10 play just tested Sonnet 5.5 again vs. Sol 6: both on high effort, same reference, same prompt both burned 1% of my weekly quota, both took ~20 mins, but sonnet's output is leagues more realistic how did OpenAi let Opus > Astra AND Sonnet > Sol happen in the same cycle? and Fable 5.5 isn't even out yet...
This post quotes another post: x.com/maxt3chno/status/2104679311663935970