Breeze TTS 2 tops open-weights TTS arena, beating Fish Audio by 90 Elo
ArtificialAnlys · x · 2026-08-26
Breeze TTS 2 from BreezeBlue (50 languages, text-prompt voice generation, streaming; weights open on Hugging Face) is now the #1 open-weights TTS model in the Artificial Analysis Provider Voices Speech Arena with an Elo of 1,215, leading Fish Audio S2 Pro (1,125) by 90 points and ranking #6 overall out of 100+ models.
- Controlled Voices: #3 among open weights (Elo 1,002), tied with Fish Audio S2 Pro and 8 Elo behind Mistral's Voxtral TTS (1,010)
- Speed: 45 chars/sec vs 102 chars/sec for Fish Audio S2 Pro
- Price: $34/1M chars on the hosted endpoint vs $15/1M for Fish Audio; both can be self-hosted
Related event: Breeze TTS 2 Tops Open-Weight TTS Leaderboard by 90 Elo(2 posts)→
More from Multimodal
- Fun images generated by GPT-Image-2 model — sloppenheimer · 2026-08-26
- ComfyUI video generation experiments: image and audio reference workflow — TensorTinkererTom · 2026-08-26
- What details instantly make you recognize an AI-generated image? — NextHeat8167 · 2026-08-26
- How to deal with 'plastic skin' in Minimax H3 ref2va generation? — Any-Scar765 · 2026-08-26
- Creator Remixes 5-Year-Old Phonk Song Using Full AI Workflow — bennash · 2026-08-26
- Using MiniMax H3 to Generate Training Data for KREA2 LoRA — NetworkSpecial3268 · 2026-08-26