GPT-5 is configured with minimal reasoning effort by default to keep latency suitable for real-time voice while still delivering high-quality conversations. The aggressively low cached input pricing of GPT-5 makes it an especially attractive choice for voice agents, where most tokens are cached input.

Groq’s gpt-oss-120b offers lightning-fast, open-source inference for those seeking performance without vendor lock-in.

You can switch models instantly and test live in the Telnyx AI Assistant Builder.

Learn more: https://telnyx.com/release-notes/gpt-5-groq-gpt-oss-120b-llm-support