llm_config block of the agent config. Each provider below has its own guide with models, configuration and troubleshooting.
Not sure which to choose? Choose providers compares them by language, latency and cost. To use your own account with a provider, see Connect provider credentials.
OpenAI
Configure GPT-5.6, GPT-5.5, GPT-5.4, or GPT-5.4-mini as the LLM for your Bolna voice agent — model selection, streaming config, and function calling.
OpenRouter
Access models from OpenAI, Anthropic, Google, Mistral, Meta, and more through a single OpenRouter API key.
Azure OpenAI
Use GPT-5.4-mini, GPT-5.4, GPT-4.1, or GPT-4o through Azure OpenAI for enterprise data residency and compliance.
DeepSeek
Reference for configuring DeepSeek models in a Bolna voice agent: V4 Flash and V4 Pro, including JSON config, prompt caching, and migration steps.
Anthropic
Reference for configuring Anthropic Claude models in a Bolna voice agent: Claude Sonnet 5, Opus 4.8, and Haiku 4.5, with JSON config and model guidance.
Google Gemini
Reference for configuring Google Gemini models in a Bolna voice agent: Gemini 2.5 Flash, Gemini 2.5 Pro, and the Gemini 3.x series, with JSON config.
Sarvam
Configure Sarvam 105B Chat, Sarvam’s conversational LLM for voice agents in 10 Indian languages and English.
AWS Bedrock
Run OpenAI’s gpt-5.6-luna on AWS Bedrock in India, so LLM requests and data stay in India.
Custom LLM inference
Use your own OpenAI-compatible model, or an inference provider such as Groq, Cerebras, Together AI, Fireworks AI, Baseten, Nebius, Simplismart or Modal, with your endpoint and your key.