Cheap TTS serving: Qwen3-TTS on vLLM-Omni at $3 per 1M characters

Melvin Vivas · X post · 2026-05-15 · Open on X

Topics: LLMOps, Deployment & Monitoring · Level: intermediate

Summary

A provider reports serving the open Qwen3-TTS model on vLLM-Omni for $3 per 1M characters. They say this is about 90% cheaper than comparable closed-source TTS APIs. They got there by optimizing a single-replica serving stack, which shows how self-hosting open models can cut costs.

Key points

• Qwen3-TTS served on vLLM-Omni costs $3 per 1M characters.
• That is about 90% cheaper than comparable closed-source TTS APIs.
• The savings came from optimizing a single-replica serving stack.
• Measure cost per concurrent stream when you size TTS serving.

Resources mentioned

• Qwen3-TTS · tool · github.com · free
  Open text-to-speech model from the Qwen family.
• vLLM-Omni · tool · github.com · free
  Version of the vLLM inference engine for serving omni-modal models such as TTS.