The amount of degradation at a given context length isn't constant though so a model with 5x the context can either be completely useless or still better depending on the strength of the models you're comparing. Gemini actually does really great in both regards (context length and quality at length) but I'm not sure what a hard numbers comparison to the latest Claude models would look like.

A good deep dive on the context scaling topic in general https://youtu.be/NHMJ9mqKeMQ