Tasks and inputs
SFX v3 is listed for Audio, Audio generation. It accepts text and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
SFX v3 is ElevenLabs' newest text-to-sound-effects model, better at multi-part prompts, music and vocals, with seamless looping and less clipping.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: Sound Effects listed at about $0.12 per minute of generated audio, not tied to a named model version. Indicative; confirm on your ElevenLabs plan.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Gongs, chimes and whistles used to clip. That small, specific fix is one of the clearest improvements in SFX v3, ElevenLabs' third-generation sound-effects model, released on 2 November 2026 under the API ID eleven_text_to_sound_v3. High-frequency sounds now come out cleaner, and the model handles a broader range of prompts than before.
loop=true for ambience beds and game backgrounds.ElevenLabs itself notes that results vary by prompt and suggests comparing v2 and v3 on your own material. That is sound advice for any generative audio tool.
The previous model, eleven_text_to_sound_v2, is still the API default. It is also deprecated, with removal set for 1 March 2027. If your code omits model_id, it is quietly using v2 today and will break or change behaviour after that date. Set the model explicitly now.
| SFX v2 | SFX v3 | |
|---|---|---|
| API ID | eleven_text_to_sound_v2 |
eleven_text_to_sound_v3 |
| Status | Deprecated, API default until removal | Current |
| Removal | 1 March 2027 | |
| Looping | loop=true |
|
| Music and vocals in prompt | Not a strength | Supported |
In ElevenLabs' creative apps (Sound Effects, Studio and Flows), v3 is already the default. It also generates the sound effects in the Video to music feature. Anyone using those tools has been hearing it since launch.
The API pricing page lists Sound Effects at about $0.12 per minute of generated audio without naming a model version. Assume it applies to v3 but confirm on your plan, and check whether short clips round up to a minimum billing unit.
Game developers needing loopable ambiences, video editors filling foley gaps, podcast and app teams producing UI sounds and stingers. It is not a music generator; for full tracks look at MiniMax Music 3 or the text-to-music models list.
Gather the last thirty sound requests your team actually made. Write each as a prompt, render with both v2 and v3, and have the person who asked choose blind. Include at least five high-pitched sounds to test the clipping fix and five loops, which you should play back-to-back for a minute to listen for seams. Track how many generations it takes to get a keeper; that number drives real cost more than the per-minute rate. More options are under audio generation models.
Compare the model on the work you need to complete.
SFX v3 is listed for Audio, Audio generation. It accepts text and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
Compare pronunciation, voice consistency and time to the first playable audio. Check the service’s billing unit for the actual script.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Alibaba
Qwen-Audio-3.1-TTS Next is Alibaba's unified audio generator: speech, multi-speaker dialogue, effects and ambience in one pass from text and timestamps.
StepFun
StepAudio 3 Gen Preview creates a complete audio scene from text, audio references and role descriptions.