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

In our September 2026 update, selected Zalify AI features moved to Claude Opus 5.5, Claude Fable 5.1, and GPT-6 Luna. The table below shows which features changed. If you use their default model settings, there is nothing to configure.

What changed

Lower base rates for some tasks

The model pricing configuration used for this update has lower input and output token rates for Opus 5.5 and GPT-6 Luna than for their predecessors:

• Claude Opus 5.5: both base rates are 20% lower than Opus 5.
• GPT-6 Luna: the input base rate is 50% lower and the output base rate is about 58% lower than GPT-5.6 Luna.
• Claude Fable 5.1: the base rates are unchanged from Fable 5.

These are comparisons of model token rates, not guaranteed savings on every completed task. The amount deducted from your Zalify balance also depends on token usage, caching, the task, and the workspace's pricing settings. Provider base rates are not the final Zalify charge. Check your usage history for actual deductions.

This update does not change image, video, or voice model pricing. See AI models and pricing for those rates.

Why different features use different models

We evaluate model changes against tasks from real stores. A model needs to perform well on the specific feature before becoming its default.

In our storefront tool-selection test, GPT-6 Luna selected the correct tool for all 123 instructions. In that test set, it also avoided starting paid image generation or rebuilding a page when the instruction did not ask for it. That result supported using it for tool selection, alongside copy edits and image prompts; it does not guarantee the same outcome for every future instruction.

Whole-page rebuilding gave us a different result. GPT-6 Luna made fewer of the requested changes than the previous model and, in one run, left the page unchanged. Whole-page edits therefore remain on GPT-5.6 Luna. We will revisit that choice when another model performs at least as well on those tasks.