Claude Sonnet 5 is the new default. And a clearer model picker.

Two changes to the model picker this month. Claude Sonnet 5is now the default model for every new chat, and the picker itself was redesigned so that choosing a model is a decision rather than a guess. OpenAI’s new GPT-5.6 family joins the lineup at the same time.

Why Sonnet 5 is the default

Most of the work in Content Chat is agentic: read twenty pages, pull Search Console, compare, decide, write. Sonnet 5 is the model that does that loop best for the credits it costs. It reasons well enough to hold a long audit together, writes in a voice that reads like a person, and moves fast enough that you are not waiting on it. It replaces Sonnet 4.6 as the starting point, and if you never touch the picker, this is the upgrade you get.

Claude Fable 5 keeps its place above it for the hardest agentic runs, and Opus 4.8 remains the heavy-analysis option. Nothing about how you switch has changed: click the model name next to the send button, pick, send.

A picker that explains itself

The list used to be names. Now each model comes with a one-line description of what it is best at and a cost-tier indicator, so you can see at a glance that a Flash model is the light, cheap pass and a flagship is the one to save for the question that matters. Legacy models are marked as such, still available for the chats that use them, but out of the way for new ones.

GPT-5.6: Sol, Terra and Luna

OpenAI’s new generation arrives as three tiers, and all three are in the picker:

• GPT-5.6 Sol - the flagship, for complex agents and the longest reasoning chains.
• GPT-5.6 Terra - the balanced choice for everyday production work, and the new default for OpenAI-based workflows.
• GPT-5.6 Luna - the fastest and most affordable tier, for well-defined tasks over many pages.

As with every model in Morrison, the same Skills, integrations and crawl are available on each. Credits scale with the model you choose, per message.

Try it

Open a new chat and Sonnet 5 is already selected. Open the picker to see the descriptions and cost tiers, and try a Flash model for a quick pass over fifty pages before handing the hard case to a flagship. Everything is live now, in every workspace.

Tell us which model earns its place in your day at hello@morrison.app.