Summary

Add xAI Text-to-Speech as a first-class TTS provider alongside ElevenLabs, OpenAI, and Edge TTS.

Why

xAI just launched their TTS API at $4.20 per 1M characters — roughly 70x cheaper than ElevenLabs API pricing. The quality is competitive, with 5 expressive voices and inline speech tags for fine-grained delivery control (pauses, whispers, laughter, emphasis, pitch/speed changes).

API Details

• Endpoint: POST https://api.x.ai/v1/tts
• Auth: Bearer token
• Docs: https://docs.x.ai/developers/model-capabilities/audio/text-to-speech
• API Ref: https://docs.x.ai/developers/rest-api-reference/inference/text-to-speech

Request format

{
  "text": "Hello world",
  "voice_id": "ara",
  "language": "en",
  "output_format": {
    "codec": "mp3",
    "sample_rate": 24000,
    "bit_rate": 128000
  }
}

Key features

• 5 voices: eve, ara, rex, sal, leo
• 20 languages + auto-detect
• Inline speech tags: [pause], [laugh], <whisper>, <emphasis>, <slow>, etc.
• Output formats: MP3, WAV, PCM, μ-law, A-law
• Sample rates up to 48kHz
• Max 15,000 chars per request

Response

Raw audio bytes in the specified format.

Suggested config

{
  messages: {
    tts: {
      provider: "xai",
      xai: {
        apiKey: "xai_api_key",
        baseUrl: "https://api.x.ai/v1",
        voiceId: "ara",
        language: "en",
        outputFormat: {
          codec: "mp3",
          sampleRate: 24000,
          bitRate: 128000
        }
      }
    }
  }
}

Context

There are several open TTS provider requests (#48366 Qwen3, #43579 Deepgram, #23080 Gemini, #46543 Chatterbox). xAI's price/quality ratio makes it particularly compelling as a default provider for cost-conscious users. The API is simple (single REST call, raw audio response) and should be straightforward to integrate following the existing provider pattern.

Related issues

• feat(tts): add Deepgram Aura as first-class TTS provider #43579 — Deepgram Aura TTS provider
• [Feature Request] Add Gemini TTS as a native TTS provider #23080 — Gemini TTS provider
• Feature Request: Add Qwen3-TTS as Local TTS Provider #48366 / Feature Request: Add Qwen3-TTS as Local TTS Provider #48370 — Qwen3-TTS provider
• Feature Request: Add Chatterbox TTS Support #46543 — Chatterbox TTS
• [Feature]: Per-Agent TTS Configuration #38673 / Per-agent TTS voice configuration #20451 — Per-agent TTS config (would benefit from more provider options)