Use the model catalog to compare specifications and choose models for agents.

Find the model catalog by navigating to Build > Models. This page lists every model available on the platform, with the specifications and pricing needed to choose the right one for each agent.

Use Search, Filter, and Sort in the toolbar to narrow the catalog.

Switch between grid and list layouts with the layout toggle.

Model cards include some basic information about the models.

• Name: The name of the model, such as Claude Haiku 4.5. Some entries also show a provider route as secondary text, such as Anthropic provided through Google.
• Context window: How much input the model can consider, such as 200K tokens or 1M tokens.
• Max output tokens: The model's response ceiling, such as 8.2K, 32K, 64K, or 128K.
• Capability icons: The input types the model supports, such as text, image, and files.
• Lifecycle badges: Retired and Deprecated labels flag models to avoid for new work.
  Note: Existing agents may continue to reference older models, but choose actively supported models for new agents and plan migrations away from anything flagged Deprecated.

Select a model to view its detailed pricing (per-token input and output rates) and full technical specifications, including modalities and attachment limits. Match the model's modalities to the inputs you promise users. For example, Claude Haiku 4.5 accepts up to 10 images of 5 MB each and up to 5 files of 20 MB each. Trimble Assist and the agent preview enforce those limits when someone uploads attachments.