customers
All customer stories
Top Voice AI companies are building with Assembly.
resources
Latest Release
Text-to-Speech API
Our first text-to-speech model, AssemblyAI-TTS-1, is built for voice agents, on live calls. It sounds natural, responds fast, and keeps callers on the line.
resources
From raw utterance to structured answer—fast open models for the cleanup pass, frontier models for the reasoning pass, one API key for both.
Fast open models for the cleanup pass, frontier models for the reasoning pass. 60 models from 7 providers behind one API key—switch by changing one string.
Your LLM calls run where your Speech-to-Text does—one less network hop on every turn, no second vendor, and one key for the whole pipeline.
Pay provider rates, not gateway rates, on global routing. Competing gateways add 5% or more to every call. No hidden fees, no minimum commitment.
36%
improvement in close rate
“The new Universal-3.5 Pro speech model from AssemblyAI is best so far in terms of accuracy, latency, and language switching.”
80%
increase in customer satisfaction
“Assembly has saved us countless hours managing models, and provided exceptional accuracy.”
36%
improvement in close rate
“The new Universal-3.5 Pro speech model from AssemblyAI is best so far in terms of accuracy, latency, and language switching.”
80%
increase in customer satisfaction
“Assembly has saved us countless hours managing models, and provided exceptional accuracy.”
The same transcript, four different jobs, four different models — all routed through one endpoint.
What is runner's knee?
Based on the transcript, runner's knee is a condition characterized by pain behind or around the kneecap. It is caused by overuse, muscle imbalance and inadequate stretching. Symptoms include pain under or around the kneecap and pain when walking.
Summarize this call in three bullet points
• Caller reports persistent knee pain after long runs. • Clinician identifies likely runner's knee from overuse and weak hip stabilizers. • Recommended next steps: rest, ice, targeted glute and quad strengthening.
Extract symptoms, conditions, and recommended treatments as JSON.
{ "condition": "runner's knee", "symptoms": ["kneecap pain", "pain when walking"], "causes": ["overuse", "muscle imbalance", "inadequate stretching"], "treatments": ["rest", "ice", "strengthening"] }
Write a 2-sentence patient follow-up email summarizing the visit.
Hi — thanks for coming in today. As we discussed, your knee pain is consistent with patellofemoral pain syndrome (runner's knee), and I'd recommend two weeks of reduced mileage along with the strengthening routine I sent over.
Ship faster, spend less on tokens, and stop losing users to provider outages.
Frontier models from OpenAI, Anthropic, and Google, plus open models we host ourselves. New ones added the day they launch.
Configure backup models per request. When a provider errors, stalls, or rate-limits, your call still goes through—no code changes.
Drop into any OpenAI SDK or HTTP client. Change a base URL and a model string, and the rest of your code keeps working.
Tool calls and schema-enforced JSON out of the box, so model responses drop straight into your app.
Cached prompts cost less and respond faster on every turn of a conversation.
EU-resident infrastructure, on by default for traffic originating in the EU, for GDPR-sensitive workloads.
Opt into zero data retention per request or project-wide, so prompts and responses are never stored.
A Business Associate Agreement is available for customers processing PHI, backed by a SOC 2 Type 2 report.
Create a free account, point your OpenAI SDK at the Gateway, and start calling any open or frontier model. Already using AssemblyAI Speech-to-Text? It's the same API key.
Start building freeNo credit card required
