← All models

端到端语音语言大模型(Flex)

Published Publisher Baidu

Overview

Baidu Flex is an end-to-end speech-language model for flexible, task-oriented real-time voice interactions. Its public beta was announced on September 16, 2026. (cloud.baidu.com)

Capabilities

  • Understands speech and produces natural-sounding speech, with more than 70 preset voices.
  • Supports server-side automatic voice activity detection and client-controlled manual audio submission.
  • Supports function calling to connect external tools and business systems, as well as information retrieval for queries.
  • Supports retrieval-augmented generation (RAG) using business knowledge bases. The launch announcement described RAG as forthcoming; Baidu’s November 9 update confirms support in the beta. (cloud.baidu.com)

Uses

Intended applications include smart speakers, in-car and smart-home interfaces, voice assistants and agents, call centers, enterprise knowledge answering, and education. (cloud.baidu.com)

Access and limitations

The API uses a real-time WebSocket protocol and is in public beta. Baidu asks interested users to submit a cooperation inquiry. External task execution requires integration with the relevant tools or business systems. (ai.baidu.com)

Pricing

Not verified as of November 16, 2026. No Flex-specific official domestic CNY or international USD input/output rates could be verified. Baidu’s domestic pricing guide lists rates for the separate Pro and Lite versions, so those rates should not be assumed to apply to Flex. (ai.baidu.com)

Official sources