Realtime TTS 1.5 Max
Realtime voice-agent candidate; add API metadata, pricing, and CodeSOTA hard-text run.
Text-to-speech · watchlist
New TTS models stay unscored until evidence catches up.
These models are tracked for upcoming CodeSOTA runs. They are not ranked against measured rows, and they do not inherit vendor MOS until a source and verification tier are attached.
Realtime TTS 1.5 Mini
Realtime voice-agent candidate from the same Inworld model family; verify latency and pricing.
Gemini 3.1 Flash TTS
Google frontier TTS model with audio tags and broad language coverage; add exact API model ID.
Eleven v3
ElevenLabs latest expressive synthesis model; needs shared-prompt samples and control-following run.
Cartesia Sonic 3.5
Stable May 2026 Cartesia snapshot; verify transcript following, language quality, and p95 latency.
OpenAI gpt-4o-mini-tts
OpenAI current TTS model for prompt-controlled realtime speech; add voices, latency, and cost.
StepAudio 2.5 TTS
Contextual TTS candidate; verify public access, cloning claims, and language coverage.
MiniMax Speech 2.8 HD
High-fidelity TTS candidate; verify emotion control, languages, and commercial terms.
OmniVoice
Explicit pitch, whisper, speed, duration, tags.
VoxCPM2
Open multilingual speech model to test for hard-text and code-switching.
Qwen3-TTS
Qwen speech generation candidate; verify language breadth and licensing.
Voxtral TTS
Open-weight TTS candidate; verify artifacts and serving path.
MOSS-TTS-Nano
Small-model edge candidate for browser/mobile deployment.
Chatterbox
Control stress test for exaggeration, CFG, temperature, reference audio.
VibeVoice
Expressive long-form/dialogue candidate; watch for drift and repetitions.
Dramabox
Narration and character-voice candidate.
Scenema
Scene/dialogue TTS candidate; needs artifact-backed eval.