You can now select your preferred providers for Large Language Models (LLM), Speech-to-Text (STT), and Text-to-Speech (TTS). Balance between latency, quality, and price based on your priorities—whether you need lightning-fast responses, the highest quality interactions, or the most cost-effective solution.

With Custom Model Selection, you can:

• Pick your LLM provider: Choose the LLM provider that best fits your needs—OpenAI for versatility, Together AI for open-source flexibility, Azure OpenAI for enterprise compliance and data residency requirements, or Oration OpenAI, which combines the best of multiple models to deliver the fastest and smartest replies.
• Select your speech recognition provider: Pick between Deepgram for ultra-low latency or Azure for enterprise-grade accuracy and integration with your existing Microsoft infrastructure.
• Customize voice output: Use Eleven Labs for natural, expressive voices in global languages (including English), or Sarvam for authentic Indian language support—ensuring your AI agents sound natural to your customers.
• Mix and match providers: Combine different providers to create the perfect stack for your specific requirements, whether that's cost optimization, latency reduction, or regional compliance.

This flexibility means you can tailor each AI agent to excel in its specific role—whether handling customer support calls in specific languages, processing sensitive data within Azure's compliance framework, or delivering the most natural-sounding voice experiences.

To get started, open any agent, navigate to the Advanced Settings tab, and select your preferred providers for LLM, STT, and TTS from the dropdown menus.