Claude Opus 5.5, Fable 5.1 and GPT-6 are now in Zalify

Anthropic and OpenAI both shipped new models this month. Zalify's AI features now use them. You don't need to do anything: the next run you start uses the new model.

What changed

Most runs get cheaper

AI usage is priced from each provider's list rate, so a cheaper model means a cheaper run:

• Claude Opus 5.5 lists at $4 / $20 per million input / output tokens, down from $5 / $25 for Opus 5. A Max-mode run costs about 20% less.
• GPT-6 Luna lists at $0.10 / $0.50, down from $0.20 / $1.20. Fast-tier runs and storefront copy edits cost less than half what they did.
• Claude Fable 5.1 costs the same as Fable 5 ($10 / $50).

Per-model image, video and voice prices are unchanged. They are on the AI model pricing pages.

How we switched

We don't swap a model just because a newer one exists. Each feature has a benchmark built from real stores, and a new model has to match the old one before it replaces it.

GPT-6 Luna routed every storefront-editor instruction in our test set to the right tool, 123 out of 123, and never started a paid image run or rebuilt a page that nobody asked for. So it takes over copy edits, image prompts and routing.

It did not win everywhere. Asked to rebuild a whole storefront page, GPT-6 Luna made far fewer changes than the model it would replace, and once left the page as it was. Whole-page edits stay on GPT-5.6 Luna for now. We'll move them when a new model does the job at least as well.