Loading...
Power your voice agents with the world's leading AI models. Choose from OpenAI GPT-4, Anthropic Claude, Google Gemini, or Groq based on your needs. No vendor lock-in.
LLM Providers
Model Options
Latency
Uptime
World-class LLMs powering natural voice conversations
Access GPT-4o, GPT-4o Mini, and GPT-4 Turbo. Best for complex reasoning, nuanced conversations, and broad knowledge.
Claude 3.5 Sonnet and Claude 3 for reliable, safe, and helpful conversations. Excellent for customer service and compliance-sensitive use cases.
Gemini Pro and Gemini Live for real-time multimodal understanding. Strong multilingual capabilities for global deployments.
Groq's LPU-powered inference for ultra-low latency. Best when response speed is critical for natural conversation flow.
Switch between models without changing your agent logic. Test different models and compare performance easily.
Different agents can use different models. Use GPT-4 for complex sales calls, GPT-4o Mini for simple reminders.
Flexibility to choose the right model for every use case
No Vendor Lock-in
Switch models anytime
Best Model per Use Case
Optimize for each scenario
Future-Proof
New models added regularly
Custom Fine-Tuning
Use your fine-tuned models
Cost Optimization
Right model, right price
Latency Control
Choose speed vs. capability
A/B Testing
Compare model performance
Fallback Support
Auto-switch if one fails
Choose the right model for your use case
| Model | Speed | Capability | Cost | Best For |
|---|---|---|---|---|
| GPT-4o Mini | Fast | Good | $ | Simple reminders, notifications |
| GPT-4o | Fast | Excellent | $$ | Most use cases, balanced choice |
| GPT-4 Turbo | Medium | Best | $$$ | Complex sales, negotiations |
| Claude 3.5 Sonnet | Fast | Excellent | $$ | Customer service, compliance |
| Gemini Pro | Fast | Good | $$ | Multilingual, Google ecosystem |
| Groq (Llama) | Ultra-Fast | Good | $ | Latency-critical applications |
Simple configuration, powerful flexibility
Create Agent
Design your voice agent's behavior
Select Model
Choose LLM based on use case
Configure Fallback
Optional backup model
Deploy & Monitor
Track performance by model
Real feedback from voice AI users
"We use GPT-4 for sales calls where quality matters and GPT-4o Mini for appointment reminders. Cut our LLM costs by 40% without sacrificing quality where it counts."
40% Cost Savings
Bangalore
VP Engineering
"Switched from GPT-4 to Claude for customer service calls. The safety and helpfulness improvements were noticeable. Love being able to experiment."
Better CSAT Scores
Mumbai
CX Lead
"Groq's speed makes conversations feel completely natural. No awkward pauses. Essential for high-volume calling where every second counts."
<100ms Latency
Delhi
Product Head
Common questions about AI model selection
It depends on your use case. GPT-4o offers the best balance of capability and speed for most use cases. Claude is excellent for customer service with its safety focus. Groq provides the lowest latency for natural conversations. We recommend starting with GPT-4o Mini for cost-effectiveness and upgrading as needed.
Yes, switching is as simple as changing a dropdown in your agent configuration. Your prompts, functions, and workflows remain the same. We handle the API differences internally so you can switch without code changes.
You pay the underlying LLM costs plus our platform fee. GPT-4o Mini is most economical, GPT-4 and Claude 3.5 are mid-tier, and GPT-4 Turbo is premium. We provide cost estimates per call based on your conversation length and model choice.
Yes, each agent can be configured with its own LLM. You might use GPT-4 for complex sales qualification calls, GPT-4o Mini for simple appointment reminders, and Claude for sensitive customer service interactions.
Voice calls need low latency for natural conversation. Groq offers the fastest inference (<100ms). GPT-4o and GPT-4o Mini are also optimized for speed (~300-500ms). We use streaming responses and voice caching to minimize perceived latency.
Yes, if you have fine-tuned models on OpenAI, you can use them with our platform. Contact us for custom model integration including self-hosted or private LLM deployments.
We support automatic fallback. If your primary model (say GPT-4) is slow or unavailable, we can automatically route to a backup (GPT-4o Mini or Claude). This ensures your calls continue without interruption.
Yes, our analytics dashboard shows performance metrics by model - conversation quality scores, call completion rates, and latency. You can run A/B tests to compare how different models perform for your specific use case.
Explore more AI Voice Assistant capabilities