# Murf Falcon: 55ms-latency model for voice agents across 35 languages

## TL;DR

[Murf launch announcement, Nov 2025]: Falcon is Murf's API built for real-time voice agents, with 55ms model latency across 35 languages and sub-130ms time-to-first-audio, priced per character. The practical takeaway: Falcon for conversational agents where latency is the product; GEN2 for batch narration where quality per minute matters more. They are different model lines with different request shapes and different voice IDs, so confirm the voice ID exists on the model you are calling (GET /speech/voices) before wiring it into an agent.

Use this when you run into the situation in the title.

**When not to use this skill:** unrelated tasks. It covers only the procedure above.

## Compatibility

The steps above apply to the commands named in them. This skill does not pin a version, so if a flag looks different on your machine, check your installed version's docs first.

## Details

Context: Web, Murf AI launch announcement (Business Wire, Nov 2025): Murf launched Falcon, a text-to-speech API built for real-time voice agents, with 55ms model latency across 35 languages, positioned against ElevenLabs. Falcon is priced per character and aimed at conversational agents; GEN2 remains the batch/narration model. The launch details matter for picking the model line: low-latency agent work goes to Falcon with its sub-130ms time-to-first-audio, long-form narration stays on GEN2.