On "ttok 1.0"

This is a tool problem dressed up as a release note, and it matters for anyone doing SEO with AI. Defaulting to outdated tokenizer versions is the kind of hidden assumption that breaks workflows quietly—you're counting tokens one way, Claude or GPT-4o is processing another, and your prompt engineering feels broken when it's just your baseline that's wrong. If you're using ttok to validate token budgets for any LLM-driven SEO work (content briefs, extraction, classification), you need the right default immediately, not six months in when you notice your estimates are systematically off by 15 percent.