Telnyx Edge Compute Is Now Available

Telnyx Edge Compute is now available. The agent runtime layer now exists on the platform where the agent's own code lives onto the same infrastructure that already hosts its inference, speech, and calls.

Developers can now run an agent's own code on the same platform that already powers its inference, speech, and telephony. Telnyx Edge Compute completes the stack for real-time voice AI agents. The runtime was the one piece that still ran a hop away on someone else's cloud. Now it's live on Telnyx.

What's New

• Functions (GA): Containerized edge compute, deployed with one command (telnyx-edge ship). Reaches every runtime component through bindings, eliminating credentials and network hops from your code.
• StatefulActor (beta): Durable per-entity state with single-threaded execution. One instance per entity, one call at a time.
• KV (GA): Globally distributed key-value storage. Accessed through bindings, with no network hops, credentials, or egress fees. Immediate consistency between binding and REST API.
• CloudFS (beta): Shared POSIX filesystem backed by Cloud Storage. Mount on any container, files persist across restarts.
• SQLDB: Create a SQLite database, bind it in your config, query it from your functions. No database server to manage, no infrastructure to provision.

These join our pre-existing products:

• Object Storage (GA): S3-compatible object storage with zero-credential access from edge functions.
• Inference (GA): Open models on owned GPUs, up to 75% less than proprietary alternatives, hosted in region on 4 continents, with zero data retention.

Why It Matters

The whole agent pipeline now runs on one network. Your agent's code, state, storage, inference, speech, and calls sit on one platform, replacing up to six vendors you would otherwise have to stitch together.

Functions reach state, storage, and inference through bindings rather than network calls, so the whole pipeline stays under 200ms round-trip. Inference runs on GPUs Telnyx owns, priced on the cost of running the model, up to 75% less than proprietary alternatives, in region by architecture. One API, one SLA, one bill, and end-to-end observability replace a stack of dashboards.

What's Coming

• WebSocket (coming soon): Live connections into StatefulActor.
• AgentSDK: Bring LangGraph, CrewAI, or the Vercel AI SDK unchanged and run them in a durable StatefulActor.
• Operating Envelope: Govern what agents can do. Developers ship the manifest, operators attach policy.
• Memory: Cross-channel conversation recall across sessions and surfaces.

Where to Test

Try the demos on our new Microsite, or explore the docs and see what you can build.