Multimodal King ProprietaryPay per Token / API

Gemini Omni Flash / Veo 3.1 (2026 Benchmarks)

By Google DeepMind · Released 2025-01

Google's advanced video model with native multimodal audio-visual generation, film-grade cinematic control, and unmatched understanding of complex storyboards.

Benchmark Scores & Technical Specifications

Arena Elo Rating1240
VBench Quality Score92.8%
Sub-Metrics (Motion / Visual / Prompt)92% / 94% / 94%
Supported Modalitiestext to video, image to video
Max Resolution & Duration1080p / 4K Ready · Up to 10s+ (30 fps)
Typical Latency~25s (Cloud TPU)
Access & PricingPay per Token / API (Integrated in Gemini API / Vertex AI)

Key Capabilities & Strengths

  • Native synchronized sound and visual consistency
  • High-speed Flash latency on Vertex AI
  • Nuanced lighting, depth-of-field, and cinematic styles

Frequently Asked Questions

What is Gemini Omni Flash / Veo 3.1's Arena Elo and VBench score?

Gemini Omni Flash / Veo 3.1 holds an Arena Elo rating of 1240 and a VBench quality score of 92.8%.

Is Gemini Omni Flash / Veo 3.1 open-weights or proprietary?

Gemini Omni Flash / Veo 3.1 by Google DeepMind is a proprietary commercial model (Pay per Token / API — Integrated in Gemini API / Vertex AI).

What resolution, clip duration, and speed does Gemini Omni Flash / Veo 3.1 support?

Gemini Omni Flash / Veo 3.1 supports up to 1080p / 4K Ready resolution, 10s+ duration at 30 fps, with typical generation latency around ~25s (Cloud TPU).

Compare Other Top Video Models

Full Leaderboard