ElevenLabs sound effects v3 and agent LLM deprecations (November 2, 2026)

What changed

ElevenLabs posted its changelog on November 2, 2026. The parts developers need to act on:

Sound effects v3. eleven_text_to_sound_v3 is available to all workspaces and is now the default in Sound Effects, Studio and Flows. ElevenLabs says it follows multi-part prompts more closely and handles music and vocals better. eleven_text_to_sound_v2 is deprecated and will be removed on March 1, 2027.
Agent LLMs. New agents now default to gemini-3.5-flash instead of gemini-2.5-flash; existing agents keep their model. gpt-6.1-terra can be picked as an agent or analysis LLM.
Deprecated agent LLMs. gpt-4.1-nano and gpt-4o-2024-05-13 are deprecated now. gpt-5, gpt-5-mini, gpt-5-nano and their 2025-08-07 snapshots are deprecated ahead of OpenAI's December 11 shutdown, with traffic moving to newer models from November 11 to December 6. Claude Sonnet 4.5 is deprecated on November 30, replaced by Sonnet 4.6. The gemini-2.5-flash and gemini-2.5-flash-lite fallback ramp now starts November 8 instead of November 1.
SDKs. JavaScript and Python SDK v2.76.0, and a first Unity SDK, v0.1.0.

Sound effects over the API: POST https://api.elevenlabs.io/v1/sound-generation still defaults to eleven_text_to_sound_v2 in the API reference. Set model_id to eleven_text_to_sound_v3 to use the new model, and do it before March 1, 2027.
Agents: open each agent's LLM setting. If it uses gpt-4.1-nano, gpt-4o-2024-05-13, a gpt-5 model or Claude Sonnet 4.5, pick a current model before the dates above.