Choose the Right AI Model for Your Agent

Available Models

This is the summary of models that will help you get the most out of your AI agent:

For the full list of available models, please check this article: Available AI models and capabilities

GPT-4.1 ⭐ Recommended Default

The best choice for most use cases. GPT-4.1 delivers reliable performance for ticket deflection, internal search, research assistance, and customer engagement.

Recommended for:

• Ticket deflection and customer support agents
• Internal knowledge base search
• Employee onboarding and training assistants
• Research assistants
• Lead engagement and pre-sales support

Start here unless you have a specific reason to choose another model.

Supported agent capabilities: Optimal, Accuracy, Understanding.

GPT-4o mini

Optimized for sub-second response times, at the cost of reduced intelligence.

Recommended for:

• Instant responses (under 1 second)
• Simple lookups and quick conversations

Trade-off: Less sophisticated than GPT-4.1. Not recommended for complex analysis or nuanced conversations. Increases hallucination likelihood.

Supported agent capabilities: Speed.

GPT-5.1

Advanced reasoning model for complex, multi-step problems.

Recommended for:

• Analysis requiring deeper insights
• Complex interpretation and cross-linking content
• Tasks requiring sophisticated reasoning

Trade-off: Slower response times than GPT-4.1. Best for quality-critical applications where users will wait a little bit more for better answers.

Supported agent capabilities: Optimal (Enterprise), Accuracy(Enterprise), Understanding(Premium & Enterprise).

GPT-5.2 (Experimental)

Advanced reasoning model for complex, multi-step problems with experimental features.

Recommended for:

• Analysis requiring deeper insights
• Complex interpretation and cross-linking content
• Tasks requiring sophisticated reasoning

Trade-off: Slower response times than GPT-4.1. Best for quality-critical applications where users will wait a little bit more for better answers.

Note: We've carefully optimized all our models to deliver the best possible performance, so you typically don't need to worry about the technical details. However, GPT-5.2 is our newest addition, and we're still fine-tuning some optimizations to ensure it meets our high standards.

Supported agent capabilities: Optimal (Enterprise), Accuracy (Enterprise), Understanding (Premium & Enterprise).

Claude Sonnet 4.5

Anthropic's advanced model combining high intelligence with natural, human-like communication.

Key advantage: Higher accuracy than GPT-4.1 on many tasks, with excellent conversational quality.

Recommended for:

• Premium customer experiences requiring empathy
• Nuanced, context-sensitive conversations
• High-quality content generation
• Sophisticated dialogue handling

Trade-off: Slightly slower than GPT-4.1. Enterprise plans only.

Supported agent capabilities: Optimal, Accuracy, Understanding.

Claude Opus 4.6

Anthropic's advanced model built for complex problem-solving and sophisticated reasoning. A strong choice for organizations that need dependable, high-accuracy responses across demanding enterprise use cases.

Recommended for:

• Mission-critical analysis and decisions
• Complex scenarios with competing priorities
• Highest possible accuracy and insight
• Tasks where other models fall short

Trade-off: Higher query cost and potential added latency. Enterprise plans only.

Supported agent capabilities: Optimal, Accuracy, Understanding.

Gemini 3 Pro

Google AI's advanced model delivering balanced performance with natural, conversational responses.

Key advantage: Alternative provider option with performance similar to Claude Sonnet 4.5. Offers provider diversity for Enterprise deployments.

Recommended for:

• Advanced customer engagement
• Nuanced, context-aware conversations
• High-quality content generation
• Sophisticated reasoning tasks

Trade-off: Latency

Supported agent capabilities: Optimal, Accuracy, Understanding

🚧

Gemini 3 Pro is still in the preview mode on Google's platform. Performance and latency may occasionally be inconsistent.

Gemini 2.5 Flash

Google AI's lightweight model optimized for faster responses.

Recommended for:

• High-volume query handling
• Quick responses where speed matters
• Straightforward information retrieval

Trade-off: Same performance trade-offs compared to other lightweight models (GPT-4o mini, GPT-4.1 mini, Claude 4.5 Haiku). Best suited for use cases where provider diversity is important. Enterprise plans only.

Supported agent capabilities: Speed, Optimal.