Notion Adds Grok 4.3 and Grok Build 0.1 to Custom Agents.

Notion has added support for two new xAI models across its agent platform: Grok 4.3 and Grok Build 0.1. The models are now available for both Notion Agent and Custom Agents, giving teams additional options for powering their automated workflows.

The announcement, shared on LinkedIn, marks the latest expansion of model options within Notion's AI infrastructure. The company has been steadily adding model support since launching Custom Agents, previously introducing Gemini 3.5 Flash alongside a model comparison tool.

Grok 4.3 represents an incremental update to xAI's model line, while Grok Build 0.1 appears to be a new variant focused on code generation and development tasks. The addition gives Notion users access to xAI's latest language models without needing to manage API keys or infrastructure separately.

Custom Agents, which Notion refined through a beta program, handle entire workflows autonomously within the platform. Teams can now select from multiple model providers when configuring agents for tasks like documentation generation, data analysis, or customer support automation.

The model selection matters because different tasks benefit from different model strengths. Some models excel at creative writing, others at structured data extraction, and still others at maintaining context over long conversations. By supporting multiple providers, Notion lets teams match model capabilities to specific workflows.

The new models integrate with Notion's Developer Platform, announced earlier this year. Developers can write custom tools that agents use to generate assets, query live data, or interact with external APIs. The platform includes Workers for syncing external data sources into Notion databases and a CLI for building agent capabilities programmatically.

The architecture means agents aren't limited to conversational tasks. They can execute code, manipulate databases, and orchestrate multi-step processes across systems. Adding new models like Grok 4.3 and Grok Build 0.1 expands the reasoning capabilities available to these programmatic workflows.

Notion's approach differs from standalone AI assistants by embedding agents directly into the workspace where teams already manage projects, documentation, and knowledge bases. The integration means agents can reference existing content, update databases, and trigger workflows without context switching.

The update comes as workspace platforms race to integrate AI capabilities. Slack has added AI-powered search and summaries, while Airtable introduced AI field types for automated data enrichment. Notion's strategy of supporting multiple model providers rather than building proprietary models distinguishes its approach.

For small teams and builders, the expanded model options provide flexibility without requiring deep AI expertise. Teams can experiment with different models to find the best fit for their use cases, then scale usage based on results. The platform handles authentication, rate limiting, and error handling.

Notion has not disclosed pricing details for the new models or whether they require a specific subscription tier. The company typically makes agent features available to paid plans, with usage-based pricing for API calls and model inference.

Developers interested in building with the new models can access documentation at developers.notion.com. The platform supports TypeScript for custom tools and provides examples for common integration patterns like syncing external data sources and generating content from templates.