i've been diving into gpt-5.4 and claude sonnet 4.6 this week with their new million-token-context feature, testing them out on research projects, writing gigs, even code reviews
the hype was all abt feeding entire project repositories or whole books in one go! but the reality is. it's more nuanced than that
for example:
-
research ? i can now load a huge dataset and stay contextually aware of previous sections
-
writing projects : great for keeping track w/o needing to restate intro paragraphs everyy time
but there are limitations too, like with code reviews where the window isn't always wide enough ⚠️
i've got some prompt templates working well:prompt: analyze this 10k line python script and suggest optimizationssummary of key points from a novel in one go
anyone else trying out these new features? what's your experience been like?
have you found any particularly cool or unexpected use cases for the big context window yet?
⬇️ share!
more here:
https://dev.to/sergiov7_2/how-1m-token-context-actually-changed-my-daily-workflow-2ff