Alibaba has been on a roll across every AI modality in 2026, and its latest move is in text-to-speech. Qwen-Audio-3.0-TTS-Plus recently claimed the top spot on the Artificial Analysis Speech Arena leaderboard for provider voices, with an Elo score of 1,236, narrowly edging out SpeechifyAI's Simba 3.2 at 1,234. The margin is razor-thin, with overlapping confidence intervals, but the result signals that Alibaba's voice AI is now competitive with the best commercial offerings in the world.

Where it stands in the field

The Speech Arena uses a blind Elo rating system: real users listen to pairs of audio clips generated from the same text and vote for whichever sounds more natural. It is a human-preference benchmark, not a lab metric, which makes it one of the most meaningful quality signals in TTS. Qwen-Audio-3.0-TTS-Plus earned its Elo score based on 1,305 arena appearances, placing it ahead of Google's Gemini 3.1 Flash TTS at 1,214 and Cartesia's Sonic 3.5 at 1,207.

By mid-2026, Alibaba has assembled one of the most extensive model lineups in the industry, with recent leading launches spanning language, image, and video generation. The TTS win is the latest chapter in that story.

The tradeoff you need to know

Quality leadership comes with a real cost: throughput. Qwen-Audio-3.0-TTS-Plus generates just 16 characters per second, compared to Simba 3.2 at 30.2, Gemini 3.1 Flash TTS at 27, and Sonic 3.5 at a blistering 120. If you are building a real-time voice assistant where latency is everything, the speed gap matters enormously.

Pricing lands in the middle of the pack at $27.59 per 1M characters via Alibaba Cloud Model Studio. That puts it below Sonic 3.5 ($39.00/1M) but above Gemini 3.1 Flash TTS ($18.31/1M) and Simba 3.2 ($10.00/1M). Here is how the top four compare at a glance:

ModelElo ScoreThroughput (chars/sec)Price (per 1M chars)
Qwen-Audio-3.0-TTS-Plus
Alpha Signal

Don't miss what's next in AI

Join 300,000+ engineers and researchers who get the signal, not the noise.

  • Full access to in-depth AI research breakdowns
  • Be the first to know what's trending before it hits mainstream
  • Daily curated papers, repos, and industry moves