Models
Explore current Specify models, plan availability, context limits, and reasoning effort.
Specify offers six models from the Anthropic Claude and OpenAI GPT families. Choose a model for each conversation in the chat message box, and set the default models for automations in workspace settings.


Compare models
| Model | Provider | Available on | Context | Max output |
|---|---|---|---|---|
| Claude Opus 5.5 | Anthropic | Pro and above | 1M | 128K |
| Claude Sonnet 5.5 | Anthropic | Pro and above | 1M | 128K |
| Claude Haiku 5.5 | Anthropic | All plans, including Free | 1M | 128K |
| GPT-6.1 Sol | OpenAI | Pro and above | 1.05M | 128K |
| GPT-6.1 Terra | OpenAI | Pro and above | 1.05M | 128K |
| GPT-6 Luna | OpenAI | All plans, including Free | 1.05M | 128K |
All supported models offer at least 1 million tokens of context: 1 million for Claude and 1.05 million for GPT. Context is the conversation and source material a request can work with; it is separate from the maximum output token count.
- For current prices and limits, see the pricing page. For how usage is calculated, see Usage and AI budget.
- Every model supports tool calls (searching knowledge, reading and writing documents, and more) and parallel tool calls.
- The default model for new conversations is Claude Haiku 5.5.
Model characteristics
Compare accuracy and speed using the same material and request. Choose a model for your task by considering both the quality of the result and usage.
For which model to use for each task, see Choosing a model.
Reasoning effort
You can set reasoning effort to one of six levels: Off, Low, Medium, High, Extra, or Max. Claude Opus 5.5 and GPT-6.1 Sol support five levels, from Low through Max. Higher effort means deeper reasoning before answering, but responses are slower and use more tokens.
On the Free plan, reasoning effort is fixed at Off. When you switch models, the current reasoning effort is adjusted to the range the new model supports.
Where to choose a model
| Location | Applies to |
|---|---|
| The model button in the chat message box | The whole current conversation. Change it after the response finishes. |
| Settings → Workspace → General → Default AI Model | Automations in this workspace |
Models that aren't available on your plan show a lock in the menu, and selecting one shows a plan upgrade prompt. The selected model is used for the request.
Understand models, skills, plugins, and connectors
After choosing a model, decide which sources it should read and how it should work. Connected sources and skills shape the result with any model.
| Component | Role | Example in Mamma (a parenting app) |
|---|---|---|
| Model | Understands your request, reasons over sources, and produces an answer. | Summarize the record-saving flow with Claude Haiku 5.5. |
| Skill | Instructions and review steps for repeatable work. Built-in skills are always on; workspace skills have usage controls. | Investigate the call flow with code analysis and save the result using document-writing instructions. |
| Plugin / integration | Makes the capabilities of external services and MCP servers available for the agent to discover and use. | Check the GitHub connection and prepare access to the repository. |
| Connector | The screen for managing service account connections and your AI access preferences. | Check Connected and AI access on on the GitHub card. |
Review skills
Open Skills & connectors from the account menu to reach Tools. In Skills, review the built-in instructions and create skills for recurring team work. When switching models, also check the skills and sources needed for the task.


Connect external sources
Find a service in Tools → Connectors. Check both Connected and AI access on. Google connects Gmail, Calendar, and Drive through one card. To let an external editor read Specify sources, follow the MCP instructions in Connect editor.


Try it with Mamma
- Choose Claude Haiku 5.5 in the chat composer.
- Check the GitHub connection and access to the Mamma repository in Connectors.
- Specify the sources and result: “Explain Mamma’s record-saving flow with code evidence.”
- Review cited files, tool results, and the saved document.
For details, see the skills guide, connectors guide, and MCP guide.