VAST Closes 3B RMB Series B, Launches Tripo P2.0 with Native Quad-Topology 3D Generation

TIQEX

VAST Closes 3B RMB Series B, Launches Tripo P2.0 with Native Quad-Topology 3D Generation
VAST Closes 3B RMB Series B, Launches Tripo P2.0 with Native Quad-Topology 3D Generation

2026-09-02 · AI Tech Insights · Curated from 爱范儿

Summary: Chinese 3D generative AI leader VAST announced the closing of its Series B and B+ funding rounds totaling approximately RMB 3 billion (about USD 446 million), led exclusively by Matrix Partners China, with strategic investments from Perfect World, 37 Interactive Entertainment, BlueFocus, Thundersoft and other industry giants. Alongside the funding, VAST launched Tripo P2.0, its flagship 3D foundation model built on the proprietary Nexus generation framework, achieving native quad-topology mesh end-to-end generation on diffusion models for the first time. The generated assets can be directly imported into mainstream engines, support animation rigging and fine editing, marking the transition of 3D generation from demo reels to production-ready pipelines.

VAST (Sanqi Wanwu) announced on September 1st the completion of its Series B and B+ funding rounds totaling approximately RMB 3 billion (about USD 446 million). The round was led exclusively by Matrix Partners China, with strategic investments from Perfect World, BlueFocus, Thundersoft, 37 Interactive Entertainment, Core Dynamic Investment, Yanqu Games, State Radio and Television Investment, and other industry leaders. Financial investors include CDH VGC, CICC Capital, CMC Capital Partners, Primavera Capital, Dachen Cai Zhi, and others. Huaxing Capital served as exclusive financial advisor. With this round, VAST has raised approximately RMB 5 billion within six months, setting a new record in China's 3D AI sector.

Alongside the funding, VAST launched Tripo P2.0, its flagship 3D foundation model built on the proprietary Nexus generation framework. P2.0 achieves native quad-topology mesh end-to-end generation on diffusion models for the first time, simultaneously meeting three industrial-grade thresholds: engine-ready, animation-riggable, and editable. It delivers comprehensive upgrades across precision, polygon count range, local editing, and semantic-level part segmentation.

VAST's algorithm team previously won the Best in Show award at SIGGRAPH 2026's Real-Time Live! session with Tripo P1.0. The Nexus framework reconstructs the paradigm of 3D mesh generation by decoupling vertex and topology generation, using hierarchical diffusion to generate vertices globally from coarse to fine, then encoding arbitrary topology through 'spatiotemporal intervals' — the first implementation of orderless native mesh generation on diffusion models.

P2.0's capabilities precisely address the core bottlenecks of the Vibe Coding era: second-level generation speed supports rapid Agent loops of 'generate—inspect—modify—regenerate,' budgeted clean topology meets asset requirements for web and real-time interactive applications, and semantic-level part segmentation enables large language models to directly identify and program each component — making Tripo P2.0 the most suitable 3D foundation model for Vibe Coding to date.

In commercial deployment, Tripo has been integrated into NetEase's Eggy Party and Yanyun Shiliu Sheng; in 3D printing, it covers Bambu Lab, Black Graph, Anycubic, Creality, and Elegoo; the overseas creative platform Intangible AI, founded by Apple and Pixar members, has integrated Tripo models. Tripo Studio platform has attracted tens of millions of global creators, ranking #1 in AI 3D model generation volume worldwide.

The world model research project Project Eden, released in June, provides the underlying engine for next-generation interactive content. Unlike conventional approaches of 'action-conditioned video generation' and 'static 3D scene generation,' Project Eden natively decouples underlying state inference from visual rendering, naturally unlocking persistent environments, scene reuse, and multi-user concurrent interaction. It has achieved end-to-end text-to-interactive-simulation-scene generation, with generated assets featuring collision bodies, joints, and physical properties that can be directly imported into Nvidia Isaac Sim and Gazebo for embodied AI training at scale.

Founded in 2023 by Simon Song and Yanpei Cao, VAST focuses on 3D-native foundation models and world models. At SIGGRAPH 2026, VAST was the only Chinese company featured in the Keynote, with 5 papers accepted in Technical Papers and the Real-Time Live! Best in Show award.

Key Takeaways

  • Series B+B+ totals RMB 3B (USD 446M), led by Matrix Partners China; cumulative funding in 6 months reaches RMB 5B, setting a sector record
  • Tripo P2.0 pioneers native quad-topology mesh end-to-end generation, meeting three industrial thresholds: engine-ready, animation-riggable, editable
  • Nexus framework decouples vertex and topology generation, achieving orderless native mesh generation on diffusion models for the first time
  • P2.0's second-level generation speed and semantic-level part segmentation make it the best 3D foundation model for Vibe Coding
  • Project Eden world model integrates with Nvidia Isaac Sim and Gazebo for embodied AI training at scale
  • Tripo Studio serves tens of millions of global creators, ranking #1 globally in AI 3D model generation volume
  • Only Chinese company featured at SIGGRAPH 2026 Keynote, with 5 papers accepted in Technical Papers

📎 This article was automatically compiled by AI from 爱范儿 (2026-09-02).
All rights belong to the original authors. Used for informational purposes only.

This article was automatically compiled by AI from 爱范儿.

Copyright Notice: This article is for informational purposes only. All rights belong to the original authors. For inquiries, contact [email protected].
Published by: Tiqex · 2026-09-02