Request: Add results for new Synthetic-hosted models

[djdembeck]

Hi team! The README currently lists results for MiniMax M2 and Kimi K2 Thinking, but the Synthetic API now hosts the updated versions: M2.5 and K2.5.

Could you add benchmark results for the following Synthetic-hosted models?

• MiniMax-M2.5
• Kimi-K2.5
• GLM-4.7-Flash
• nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4

Thanks!