Blog
Longer pieces on how Claude pricing and behaviour actually change over time, what we could and couldn't confirm while building this site's facts module, and the reasoning behind decisions the calculators make silently. If a how-to guide answers "how do I do X", the posts here tend to answer "why does the site say Y" — the migration history, the pricing mechanics, and the sourcing discipline behind every number published elsewhere on ClaudeHow.
New posts appear as a model generation ships, a pricing mechanic turns out to be commonly misunderstood, or a reader's question keeps recurring in the contact inbox in a way that a single how-to page doesn't fully answer. Nothing here restates Anthropic's own release notes — every post either explains a mechanic worked through with real numbers, or documents something we specifically could not confirm and why that's worth knowing.
Why no vendor will publish the subscription-vs-API comparison
The price of a token has stopped meaning what you think it means
What we could not confirm, and why that matters
A running table of Claude pricing changes
The tokenizer mistake almost every cost estimate makes
Reading Anthropic's model deprecation schedule like an engineer
Effort is the cost lever nobody tunes
What changes when you move from budget_tokens to effort
Prompt caching is not free — here's the math
Why cache minimums are not monotonic, and why it matters
The Batch API: the discount almost nobody uses
Building this site: how we keep a cost calculator honest
The difference between Fable 5 and Mythos 5
What a 1M-token context window actually buys you
How we source every number on this site