HeyGen integrates Sora 2 and Google Veo to automate video workflows

HeyGen is shifting from a niche avatar tool to a comprehensive production hub by abstracting the complexity of state-of-the-art models like Sora 2 and Google Veo. For the streaming industry, this lowers the barrier for high-volume localized content, enabling enterprises to scale personalized marketing and training videos without a linear increase in staffing or studio costs. It signifies a move toward 'modular' video production where third-party AI models serve as specific layers—visual, vocal, and structural—optimized through a single interface. Watch for the emergence of 'Video Agents' as the standard for corporate communication, potentially displacing traditional low-budget video agencies.

The expansion comes as HeyGen experiences significant commercial momentum, reportedly reaching $100 million in annual recurring revenue (ARR) by October 2025, according to getlatka.com and sacra.com. This growth followed a $60 million Series A funding round in mid-2024 led by Benchmark, which valued the company at $500 million. Per Bloomberg, this funding was intended to fuel the development of more realistic avatars as the market for generative video software is projected to reach $18.6 billion by the end of 2026, according to analysis from vivideo.ai.

HeyGen’s integration of external models like Sora 2 and Google Veo 3.1 reflects a broader industry trend toward multimodal production hubs. For instance, ElevenLabs expanded its own Studio platform in November 2025 to include a unified timeline editor that similarly incorporates OpenAI and Google video models alongside its native audio tools. While HeyGen remains dominant in the business avatar space, it faces intensified competition from Synthesia, which per pctechmag.com (May 2026), remains the segment leader for highly structured corporate training and internal communications due to its deep integrations with learning management systems like SAP SuccessFactors. To maintain its edge, HeyGen released Avatar V in April 2026, marketed as the world’s most realistic avatar model, featuring a directable API that allows developers to program specific posturing and gaze for five cents per second of generated video.