Two AI labs released new models this week at lower prices than their predecessors. Anthropic says Opus 5.5 is about 40% cheaper per task than Opus 5, and GPT-6 Sol and Luna are listed at lower token prices than GPT-5.6. None of the cited announcements mentions changes to Shopify apps that have built-in AI.

  • Opus 5.5. Anthropic says the model is about 30% faster and 40% cheaper per task than Opus 5 (@ClaudeDevs). Cursor added it as the new top model on CursorBench at 57.8% (@cursor_ai). It also went live in VS Code with GitHub Copilot (@code) and in Devin, where Cognition reports it took first place on FrontierCode 1.1 "at a fraction of the cost" (@cognition).
  • GPT-6 Sol and Luna. Codex's release notes split the pair by job: Sol for complex coding and agent work, Luna for focused, high-volume work, both at lower token prices than GPT-5.6 (@CodexReleases). OpenAI rolled them into Codex and ChatGPT Work for paid plans and into the API, and Free and Go users can try Luna in the desktop app (@OpenAI). Devin lists Sol and Luna as well (@dabit3).
  • Why price is the headline. Two frontier labs shipped on the same day, and both led with cost rather than scores alone. They measure it differently, though: cost per task for Opus 5.5 and cost per token for Sol and Luna, so the figures are not directly comparable. Neither announcement gives dollar prices. Separately, @mattyp says tokens now cost 7% less in Cursor because of improvements to how the tool runs the model.
  • Where sellers feel it. The new models reach stores first through tools owners use directly. One seller, @gregcangialosi, built a custom Shopify theme with 234 products solo in Codex. Apps with built-in AI choose their own models, and there is no public information yet on when they will switch.
  • What to do this week. If you use ChatGPT, Codex or Claude directly, run one routine task, such as ten product descriptions, on your current model and the new one, and compare time, cost and edits. If you keep a standing instruction file, @RLanceMartin recommends auditing it for Opus 5.5, since older patterns can "hobble frontier models."

What exactly changed this week?

Anthropic released Opus 5.5, which it says costs about 40% less per task than Opus 5, and OpenAI released GPT-6 Sol and Luna, which Codex's release notes list at lower token prices than GPT-5.6. Coding tools such as Cursor, GitHub Copilot and Devin added Opus 5.5 within hours.

Will the AI inside my Shopify apps change too?

Possibly, but not automatically. Each app decides which model it uses and what it charges, and none of the cited announcements names a Shopify app that has adopted the new models.

Is it worth switching now?

It is worth testing now. Lower list prices do not show how much editing the output needs for your store, and there are no public figures on that yet, so a side-by-side on one real task gives a better answer than the launch numbers.