Your models. Your keys. One integration.

05 · Providers

Your models. Your keys. One integration.

We sell no LLM, no ASR, no TTS — we make all of them interchangeable.

OpenAIAnthropicGoogleGroqCerebrasMistralDeepSeekDeepgramAssemblyAISonioxElevenLabsCartesia

One alias table

Every LLM sits behind an OpenAI-compatible alias — OpenAI, Anthropic, Gemini, Groq, Cerebras, Mistral, Together, Fireworks, OpenRouter, DeepSeek, Perplexity, xAI — plus a generic escape hatch for anything else.

Streaming speech, both directions

ASR: Deepgram, AssemblyAI, Soniox, Gladia, Speechmatics, Cartesia and more. TTS: Deepgram Aura, Cartesia, ElevenLabs, Rime, LMNT. Batch transcription covers Groq, OpenAI, Voxtral and Wizper.

Realtime speech-to-speech too

OpenAI Realtime, Grok realtime, and Qwen Omni self-hosted — the only S2S catalog that includes a model you can run on your own GPUs.

Sealed at rest, per tenant

Provider credentials are tenant-sealed. Your keys never appear in our logs, our bills, or anyone else’s tenant.

Swap in production

Change the alias, keep the agent. Live-verified providers are marked in the portal, and evals let you A/B a swap before callers hear it.

0LLM providers

0streaming ASR engines

0streaming TTS voices