Runway Claims Top Spot in Video AI with Gen 4.5
AI startup Runway announced Gen 4.5, its latest text-to-video generation model, on December 1, 2025. The model has claimed the number one spot on the Video Arena leaderboard, maintained by independent AI benchmarking company Artificial Analysis, outperforming similar models from Google and OpenAI.
Key Features of Gen 4.5
Unprecedented Accuracy
According to Runway, Gen 4.5 can generate “cinematic and highly realistic outputs” with “unprecedented” accuracy in responding to text prompts. The model produces more consistent visual outputs even for complex prompts.
Advanced Physics Simulation
A major advancement in Gen 4.5 is improved modeling of physical interactions. Motion and object behavior are more realistic, enabling truly cinematic output quality.
Versatile Style Support
The model handles various visual styles including photorealistic, stylized, and cinematic, producing more consistent outputs across different creative directions.
Remaining Challenges
However, Gen 4.5 still exhibits logic flaws common to current text-to-video models:
- Objects disappearing suddenly
- Generation of unrealistic actions
- Consistency issues in complex scenes
Runway plans to address these issues in future updates.
Intensifying Competition in AI Video
Comparison with Major Players
The release of Gen 4.5 intensifies competition with:
- OpenAI Sora 2: Video generation model released in late 2024
- Google Veo: Google’s latest video generation technology
- Meta Movie Gen: Meta’s entry into video generation
Startup Success Story
The fact that Runway, which lacks the financial resources of major tech companies, has claimed the top benchmark position demonstrates the vibrant competition in the AI video generation space.
Industry Impact
Gen 4.5 is expected to further accelerate adoption of AI video tools among media professionals. It’s also likely to positively impact Runway’s valuation and customer acquisition.
Looking Ahead
Runway plans to continue improving the model and addressing remaining logic flaws. AI video generation technology is evolving rapidly, with even higher quality video generation expected by 2026.