It really is 5.6 sol
I’ll never say again that Microsoft’s is GPT 5.6 Flash lite
1 Like
The full-power version?
That should be pure SOL.
But how much the mindset gets in the way is another matter.
Even if the low-tier one is pure, Luna still makes money. Hurry up and update it for compatibility! 
The model name is still “Gpt_5_6_Reasoning”
No update needed
Then why did they make the pelican riding a bicycle so badly?
Could it be that sol wasn’t actually that good to begin with,
but everyone used it to test models’ capabilities,
and then OpenAI specifically trained for it? 
Or is it simply that MS set the reasoning level low?
Just curious when the next version will be updated, haha.
Surely Sol can’t do an even worse job than Doubao, right?
I feel like Ruanzi itself just isn’t very good at making tool calls during conversations.