X · @teortaxesTex
· X / Twitter
One thing I truly despise is Y axis shenanigans what happened at CAISI between May and today? V4 was at 800 when evaluated, on par with GPT-5. Now it'…
One thing I truly despise is Y axis shenaniganswhat happened at CAISI between May and today? V4 was at 800 when evaluated, on par with GPT-5. Now it's equal to 5.2 (both GLM and GPT), and around 1200? Everything else also moved up? Did they change the benchmark mix?Samuel Hammond 🦉: New CAISI eval finds GLM-5.2 has the