UPDATE: I've been researching this all day and the biggest drain has been having long conversations iterating with artifacts which takes up A LOT of usage. It told me to create and edit section by section and start new chats frequently. Also, I was using tons of screenshots and so we've decided that I should use different tools to get text descriptions of my screenshots then use that in Claude. From Claude: The Reality of Images in Claude Images are expensive in tokens. A single screenshot can cost 1,000–4,000 tokens depending on resolution. If you're uploading 5–10 screenshots for mood boarding or design inspo in one session, that's potentially 20,000–40,000 tokens before you've typed a word. For that use case, claude.ai is genuinely not the right tool. Your Options ChatGPT Plus ($20/month) Your instinct is sound. For high-volume image analysis — screenshots, design inspo, aesthetic extraction — the flat rate makes much more sense than paying per token. GPT-4o's vision is excellent at describing visual style, layout, color palettes, typography feel, and design language. This is a legitimate and smart split-tool workflow. Google Gemini Gemini 2.0 Flash in particular is very strong at vision tasks and is extremely cheap via API, or free at generous limits on gemini.google.com. Worth considering as a free alternative before paying for ChatGPT. Dedicated vision/description tools If all you need is "describe this image's aesthetic," even lighter tools work. Napkin.ai, or simply GPT-4o via the ChatGPT interface, are purpose-fit for this. @Evelyn Weiss @Julia Schürer @Mary Schiller

@Mary Schiller my pleasure 🌼

@Gabriel Omat this is so pretty!

@Marianne Vahl I appreciate that 😊

3! Fell in love with Claude Artifacts (and how easy they are to create) after I saw you using them 🥰 now I’m all in Claude. Thanks for all your education on this Evelyn. This is a gamechanger for my business