Tag: claude opus

2 articles

html vs markdownclaude tokensclaude opusai input formattoken cost testweb2mdai workflow

HTML vs Markdown for Claude: Token Test Results from 12 Real Webpages (2026)

I ran 12 real webpages through Claude in both HTML and Markdown. Same prompt, same model. Token counts and answer quality differ more than most claim.

2026-06-046 min read

claudeclaude opusclaude 4.71m contextlong contextcontext windowai researchresearch workflowweb2mdanthropic

How to Actually Fill Claude's 1M Context Window (Without Copy-Pasting 200 Webpages)

Claude Opus 4.7 has a 1M token context — roughly 200 long articles. The bottleneck isn't the model; it's how to get 200 articles into one prompt.

2026-05-247 min read