Fish Audio releases Drama 3 TTS model with open weights, tops open-source arenas

On October 29, 2026, the evaluation firm Artificial Analysis announced on X that Fish Audio had released the speech model Drama 3, and shared its own measurements of the system. Artificial Analysis reported that Drama 3 is the top open weights model on both Speech Arenas it runs, and that it beat every model tested on the Pronunciation Robustness benchmark. Within the overall rankings it placed 6th on the Provider Voice Arena and 12th on the Controlled Voice Arena. Artificial Analysis also published performance figures: Drama 3 generates 66.6 characters per second, ahead of its own Drama 3 Preview at 60.7 and Murf's Speech Gen 2 at 63.0, and behind ElevenLabs' Eleven v4 at 70.0 and v3 Conversational at 66.8. At $15 per 1M characters it is priced below Gemini 3.8 Flash TTS, which costs $16.5 per 1M. According to Artificial Analysis, Drama 3 is open weights under the Fish Audio Community License, so users can download it themselves, and the model has been added to Artificial Analysis's Text to Speech leaderboards and to the Speech Arena voting. That is as far as the available material goes.