Agent
Model Configurations
Pick the right brain and voice for the job, then tune latency, cost, and creativity.
LLM models
| Model | Best for | Latency | Cost |
|---|---|---|---|
| Claude 3.5 Sonnet | Complex reasoning, long context | Medium | Higher |
| GPT-4o | General purpose, fast responses | Low | Medium |
| GPT-4o Mini | Simple agents, high volume | Very low | Low |
Voice models
| Provider | Voices | Best for |
|---|---|---|
| ElevenLabs | 50+ | Most natural, wide emotional range |
| PlayHT | 30+ | Fast and cost-effective at volume |
Temperature
- 0.0 — Deterministic — same input, same output. Good for structured calls.
- 0.7 — Creative and varied. Good for conversational agents.
- 1.0 — Very creative. Use sparingly.
Max tokens
Default is 150. Increase it for agents that give long explanations, decrease it for simple booking agents that should stay brief.