Unsupported language code en-GB for Gemini 2.5 Native Audio — documentation location & UK accent roadmap?

Gemini API
1

Hi team,

I’m running into the following error when using the Live Audio API with Gemini Native Audio:

google.genai.errors.APIError: 1007 None. Unsupported language code ‘en-GB’
for model models/gemini-2.5-flash-native-audio-preview-12-2025

This happens when setting language_code="en-GB" in SpeechConfig, for example:

async with client.aio.live.connect(
    model="gemini-2.5-flash-native-audio-preview-12-2025", 
    config=types.LiveConnectConfig(
        response_modalities=["AUDIO"],
        speech_config=types.SpeechConfig(
            voice_config=types.VoiceConfig(
                prebuilt_voice_config=types.PrebuiltVoiceConfig(
                    voice_name="Kore"
                )
            ),
            language_code="en-GB",
        ),
        input_audio_transcription=types.AudioTranscriptionConfig(),
        output_audio_transcription=types.AudioTranscriptionConfig(),
    )
):
    ...

Questions

1. Why is en-GB unsupported for this model?

Is this a temporary limitation of the Native Audio preview, or a deliberate restriction tied to voice availability?

2. Where is this documented?

I’ve struggled to find a definitive list of supported language_code values for Native Audio models.

Is there an official page or reference that documents which language/locale codes are valid per model?

3. Is there a roadmap for British / UK accents?

I understand that it’s possible to request tone or accent using natural language (e.g. “use a British accent”), but in practice this feels:

  • inconsistent between runs

  • sometimes ignored entirely

  • prone to defaulting back to a US accent

For production voice applications, having a deterministic, config-level way to select a British accent (e.g. en-GB) is really important.

If there’s an ETA, roadmap, or recommended workaround beyond prompt-based accent steering, I’d really appreciate the guidance.

1 Like
2

Yeah I have the same issue, would love to know if there is a fix

3

I also support this request. It is also quite important for us