1. What Large Language Models (LLMs) does Telnyx use for its AI Assistant?
Telnyx is model-agnostic, allowing customers to choose the LLM that best suits their application, performance and cost requirements.
Through our AI Inference platform, customers can access a wide range of leading models including OpenAI, Anthropic Claude, Google Gemini, Meta Llama, Qwen, Moonshot, DeepSeek, Mistral and many others. Customers are not locked into a single provider and can switch models as their requirements evolve.
Useful links:
2. Are calls recorded? If so, are they stored in Australia or overseas?
Call recording is optional and fully configurable.
For Australian deployments, Telnyx supports Australian data sovereignty with Voice AI processing and storage within our Sydney infrastructure where required. Alternatively, customers can choose to disable recording entirely or configure recordings to be stored directly in their own AWS S3 or Google Cloud Storage buckets to meet internal compliance requirements.
Useful links:
3. Has Telnyx completed penetration testing of its AI platform? Is documentation available?
Yes.
Telnyx undertakes regular independent penetration testing and maintains a comprehensive security program across its platform.
As part of enterprise security reviews, we can provide additional security documentation (including penetration testing summaries where appropriate), typically under NDA.
Useful links:
4. Does Telnyx have Cyber Security certifications such as ISO 27001?
Yes.
Telnyx maintains a number of internationally recognised security certifications and compliance programs, including:
These certifications demonstrate that Telnyx follows industry best practices for information security and operational controls.
Useful links:
5. Is there a limitation on Speech-to-Text transcription? For example, can it transcribe conversations longer than one hour?
There is no practical one-hour limitation.
Telnyx Speech-to-Text is designed for real-time and long-duration conversations and continuously streams transcription while the call remains active. It is commonly used for contact centres, Voice AI agents and other applications involving extended conversations.
Useful links: