Two Ways To Advantage Of Claude Sonnet 4s 1m Context Window In Cline
Source
cline.bot
Author
Cline evaluation and engineering blog
Date
Why it matters
Prompts above 200K tokens double input cost to $6 per million and need Anthropic Tier 4 access. Plan deeper sessions and longer dev cycles knowing the price jump.
Key takeaways · AI-distilled
Cline's Nick Baumann suggests using the 1M window for longer Plan-mode conversations: load full project context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition →, settle architecture and edge cases, then switch to Act mode with that context still intact.
The post argues build-test-debug-refactor loops no longer have to restart at the old 200K limit, so Cline keeps the record of what was tried, what failed and why an approach was chosen.
Prompts above 200K tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → cost $6 input and $22.50 output per million tokens versus $3 and $15 below it; Anthropic's prompt cachingReusing the model's processed form of a repeated prompt prefix so subsequent calls skip re-reading it, cutting cost and latency substantially.Full definition → and batch processing are named as ways to offset the cost.
Baumann frames the change as asking what relevant context could help rather than the minimum needed, while cautioning that this does not mean loading the whole codebase into every conversation.
Terms in this piece · Glossary
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
prompt caching — Reusing the model's processed form of a repeated prompt prefix so subsequent calls skip re-reading it, cutting cost and latency substantially.
