Qwen-Audio-3.1-TTS-Plus
Official site ↗
API ~$19/1M chars
+ Cheapest API among top-3 arena models
− Coarser emotion tagging than Eleven
MODEL TOOL
Voice & TTS：Balanced
Voiceover, audio content, voice cloning — ranked against the Artificial Analysis Speech Arena.
1 · What are you doing?
2 · Budget tier
Recommended for · Voice & TTS · Balanced
Great Chinese naturalness, fast cloning, lower API cost.
Official site ↗
Gemini 3.8 Flash TTS
Official site ↗
API ~$16.5/1M chars
+ Drops into Google pipelines at near-zero cost
− Limited voice roster
MiniMax Speech-2.8
Official site ↗
API per character; free quota
+ Excellent Chinese naturalness and emotion transfer; widely used in voice agents
− English delivery a notch below
Video generationVeo 3.1 Fast / Runway Gen-4.5 / Midjourney Video / grok-imagine-video / Hailuo 2.3 / Vidu Q3Image generationNano Banana 2（Gemini 3.1 Flash Image） / MAI-Image-2.6 / grok-imagine-image-2.0 / Seedream 4.5 / Reve 2.1 / MidjourneyVoice & TTSQwen-Audio-3.1-TTS-Plus / Gemini 3.8 Flash TTS / MiniMax Speech-2.8AI avatarD-ID Agents / Hedra / Vidnoz3D generationHunyuan3D 3.1 / Hi3D v3.0 / Tripo P1Music generationSuno v6-mini / Eleven Music v2.5 / Stable Audio 3
How to read this
Three tiers: flagship for quality-critical work, balanced for daily production, budget for bulk and experiments. Names track the current mainstream options — the landscape moves fast, so treat the tier logic (not the exact model list) as the durable part.
Rankings and prices checked — leaderboards move weekly; always confirm pricing on the official site before committing.
Sources: Artificial Analysis · LMArena · Crafiq 3D · Taxon3D