Step TTS 2

Step TTS 2 is StepFun's dedicated text-to-speech model, converting written text into synthesized audio output. It is positioned for voice-generation workloads where natural-sounding speech and expressive prosody matter, such as narration, virtual assistants, and accessibility use cases. Third-party catalog metadata lists the model as active, with a canonical identifier used across aggregators for tracking and comparison.

Independent benchmark listings give Step TTS 2 an Elo rating of 1138 based on 165 comparisons, providing a community-style quality signal relative to other speech systems, though it sits in a middle range rather than at the leaderboard top. Aggregator pages indicate that full API provider rollout was still in progress, with pricing and endpoint availability expected to surface once deployment completes. In practice, the model fits teams seeking a StepFun-backed TTS option whose maturity and routing surface are still expanding.

StepFun (China)step-tts-2step

Open provider page Provider docs

Quick Info

Provider: StepFun (China)
Model key: step-tts-2
Release date: Mar 1, 2026
Last updated: Jul 2, 2026
Input modalities
Output modalities
Capabilities: Base catalog fields only

Limits

Output tokens: 0 tokens
Context window: 0 tokens

Latest news about Step TTS 2

StepFun (China)

CoverageAnalysis

Artificial Analysis has published a dedicated model page for Step TTS 2 (Mar 2026), profiling StepFun's text-to-speech model with quality Elo, speed, and price analysis. The page explicitly identifies the variant as Step TTS 2 (Mar 2026) under the StepFun family and positions it within a field of 95 TTS models tracked by the Speech Arena. It is presented as a StepFun-originated model entry on a third-party evaluation platform rather than an inference-host or reseller listing. The supplied excerpt reports an Arena Preference Elo of 1130 for Step TTS 2 (Mar 2026) and references provider voice and controlled voice leaderboards for comparative context. Specific price-per-1M-characters and characters-per-second throughput values are referenced as tracked metrics but not visible in the truncated excerpt. The page functions as independent third-party benchmark coverage of the exact March 2026 Step TTS 2 release, not a release announcement or capability update from StepFun itself.