New reasoning capabilities: o3-mini now available in Dust

We've integrated OpenAI's latest reasoning model, o3-mini into Dust as a capability for asssistants.

Why as a capability?

While these models excel at deep reasoning, they're not optimal as standalone conversational agents. That's why we've made O3-mini available primarily as a capability that your existing agents can leverage when needed.

Learn more about our approach to reasoning models.

What this means:

• When enabled, your agents can tap into o3-mini's reasoning capabilities for complex problems.
• You get the best of both worlds: efficient task handling overall and deep analysis when needed.

Using o3-mini in Dust:

• As the @o3-mini (using high-reasoning mode) from the chat bar.
• As a reasoning capability for your agent in the agent builder in Tool & Data sources.
• o3-mini is available in two modes: standard and high-reasoning.

To select your preferred reasoning model:

1. Enable the reasoning capability
2. Open Advanced Settings
3. Choose your reasoning model

o3-mini joins DeepSeek R1 in our growing suite of available reasoning models, reflecting our commitment to making advanced AI capabilities accessible and practical for everyday work.

The Team at Dust

PS: Want to compare different reasoning approaches? Try the same complex task with both o3-mini and R1 - we'd love to hear your findings.