
Stepfun
Stepfun is the Chinese AI frontier lab behind Step-1, Step-2, and Step-Audio models. ~$1B+ raised; multimodal foundation models. Tencent-backed.

Overview
Stepfun
Stepfun is the Chinese AI frontier lab building the Step series of multimodal foundation models — Step-1, Step-2 (multi-trillion-parameter MoE), and Step-Audio for voice tasks. Stepfun was founded by Jiang Daxin (ex-CEO of Microsoft Research Asia) and has raised approximately $1B across rounds, making Stepfun one of the most-funded but least-Western-press-covered Chinese frontier labs alongside Doubao, DeepSeek, Moonshot, and Zhipu.
Production credibility: Approximately $1B+ raised across rounds at multi-billion dollar valuation. Founded 2023 by Jiang Daxin (former CEO of Microsoft Research Asia). Shanghai HQ. Step-2 is a multi-trillion-parameter mixture-of-experts model competitive with GPT-4 and Claude on Chinese-language benchmarks. Investors include Tencent, Wuyuan Capital, and other Chinese strategic and financial investors.
Key Features
- Step-1, Step-2, and Step-Audio model series
- Step-2 — multi-trillion-parameter mixture-of-experts model, competitive on Chinese benchmarks
- Step-Audio for voice tasks including TTS and speech understanding
- Founded 2023 by Jiang Daxin (former CEO of Microsoft Research Asia)
- Approximately $1B+ raised; multi-billion-dollar valuation
- Investors include Tencent, Wuyuan Capital, and other Chinese strategic investors
- Shanghai HQ; consumer + enterprise API products including 跃问 (Yuewen) AI assistant
Ideal Use Case
Chinese enterprises and consumer-app builders that need foundation-model API access optimized for Chinese-language tasks. Also AI researchers tracking the Chinese frontier-lab landscape alongside DeepSeek, Doubao, Moonshot (Kimi), and Zhipu.
How Stepfun differentiates
DeepSeek dominates Western coverage of Chinese frontier labs due to the R1 reasoning-model release. Doubao (ByteDance) and Kimi (Moonshot) dominate Chinese consumer market share. Stepfun's positioning is the enterprise + technical-research frontier — Step-2's MoE architecture is among the largest in production globally, Step-Audio targets the voice-task segment where DeepSeek doesn't compete, and the team pedigree (Jiang Daxin, ex-MSRA CEO) gives Stepfun credibility with Chinese enterprise buyers. Under-covered in Western press relative to the funding and technical scale.
FAQ
Q: What is Stepfun? A: Stepfun is a Chinese AI frontier lab building the Step series of multimodal foundation models — Step-1, Step-2 (multi-trillion-parameter MoE), and Step-Audio for voice tasks.
Q: Who founded Stepfun? A: Jiang Daxin, the former CEO of Microsoft Research Asia, founded Stepfun in 2023. The company is headquartered in Shanghai.
Q: How much has Stepfun raised? A: Approximately $1B+ across rounds at multi-billion dollar valuation. Investors include Tencent, Wuyuan Capital, and other Chinese strategic investors.
Q: Stepfun vs DeepSeek vs Doubao vs Kimi? A: DeepSeek leads Western coverage with the R1 reasoning model. Doubao (ByteDance) and Kimi (Moonshot) dominate Chinese consumer market share. Stepfun's positioning is enterprise + technical frontier — Step-2's MoE scale, Step-Audio's voice focus, and Jiang Daxin's MSRA pedigree.
Q: Does Stepfun have a consumer product? A: Yes — Stepfun operates 跃问 (Yuewen), a consumer AI assistant similar to Doubao and Kimi, alongside its enterprise API offerings.
tl;dr
Stepfun is the Chinese AI frontier lab building Step-1, Step-2, and Step-Audio multimodal models. Founded by ex-MSRA CEO Jiang Daxin; $1B+ raised at multi-billion valuation. Tencent, Wuyuan Capital-backed. Under-covered in Western press relative to scale; competes with DeepSeek, Doubao (ByteDance), and Kimi (Moonshot).
Related
Looking for more options? Browse the AI/ML Models directory or read our best AI models listicle. Stepfun is also tracked on Crunchbase.
Why Use Stepfun

User Reviews
Similar Tools




