Models
OfficeLabs lets you choose the AI model that answers in the sidebar. Pick it from the model menu in the sidebar. Your choice applies to every conversation and stays selected until you change it.
Every model below is available on both plans and is included in your plan: requests draw on your monthly allowance, whichever model you use. When the allowance is used up, requests are billed per use if you turned on overage, and stop until it renews if you did not.
Inline suggestions in Writer do not use these models. They run on your computer, on a model that ships with the app, so they need no sign-in and do not draw on your allowance.
| Model | Made by | Context window | Reads images | Good for |
|---|---|---|---|---|
| Claude Fable 5.1 | Anthropic | 200 000 tokens | Yes | The hardest analysis and writing tasks |
| Claude Opus 5.5 | Anthropic | 1 million tokens | Yes | Long documents and complex multi-step work |
| Claude Sonnet 5.5 | Anthropic | 200 000 tokens | Yes | Everyday writing, editing and spreadsheet work |
| Kimi K3 | Moonshot AI | 256 000 tokens | No | Text and spreadsheet tasks with long context |
| GPT-5.6 Luna | OpenAI | 400 000 tokens | Yes | Quick, light tasks |
| GPT-5.6 Terra | OpenAI | 400 000 tokens | Yes | General document work |
| GPT-6 Astra | OpenAI | Not published yet | Yes | The hardest reasoning tasks |
| GLM 5.3 | Z.ai | 128 000 tokens | No | Everyday text tasks |
This page and the model pages are generated from the OfficeLabs model catalogue. See How these docs stay current.