A wave of new AI models just launched. Here's a simple guide to what's new and when to use each one.

GPT-5.6 Sol (OpenAI) — the heavy hitter. Best for hard thinking: complex problems, coding, and science. Reach for it when getting the right answer matters most.

GPT-5.6 Terra (OpenAI) — the everyday workhorse. Great for normal work tasks: writing, drafting, and general help. A solid default for most jobs.

GPT-5.6 Luna (OpenAI) — the budget option. Best for simple, high-volume tasks like sorting, tagging, or quick first drafts. Cheap and fast.

Grok 4.5 (xAI, Elon Musk's company) — a strong all-rounder, especially good at coding and getting tasks done on its own. Handy if you already use X or Grok.

Claude Sonnet 5 (Anthropic) — a well-balanced pick for writing, analysis, and everyday tasks. It's the free default in Claude, so it's easy to try.

The simple takeaway: match the model to the job. Use a big model (Sol) for hard problems, an everyday model (Terra, Sonnet 5) for regular work, and a budget model (Luna) for simple, repetitive tasks. You don't need the most powerful tool for every job — you need the right one.