Google Opens Veo Image-to-Video Across 12 Model Variants

The documentation formalizes image-to-video as a production-ready capability across the full Veo model family, giving enterprise teams a single API surface for first-frame video generation with configurable output parameters. With OpenAI discontinuing Sora (consumer access ended April 26, 2026; API shutting down September 24, 2026, per NBC News, March 2026) and ByteDance freezing Seedance 2.0 internationally, Veo faces a contracting competitive field — though Alibaba's HappyHorse 1.1, released June 22, 2026 per VentureBeat, now ranks #2 on the Artificial Analysis Video Arena. Watch whether Google Cloud customers migrate video generation workloads onto Veo APIs at scale before the Sora API sunset in September 2026.

Veo 3.1 GA models (veo-3.1-generate-001 and veo-3.1-fast-generate-001) launched November 17, 2025, per Google Cloud model documentation. On January 13, 2026, Google's DeepMind blog announced Veo 3.1 "Ingredients to Video" updates including native vertical (9:16) output, 1080p and 4K upscaling, and improved character and background consistency across scenes. Veo 3.1 Lite (veo-3.1-lite-generate-001), a lower-cost tier targeting high-volume workflows, entered preview on April 2, 2026, per Google Cloud documentation.

The competitive landscape has contracted sharply. OpenAI announced Sora's discontinuation on March 24, 2026, per NBC News, with consumer access ending April 26 and the API scheduled to shut down September 24, 2026. The shutdown followed Sora generating approximately $2.1 million in lifetime revenue against operating costs of roughly $1 million per day, per VentureBeat, June 2026. A preliminary $1 billion Disney-OpenAI deal to license characters for Sora collapsed with the shutdown, with Disney stating it "respects OpenAI's decision to exit the video generation business," per Al Jazeera, March 2026. Separately, ByteDance indefinitely postponed the international rollout of Seedance 2.0 after copyright complaints from Netflix, Warner Bros., Disney, Paramount, and Sony, per VentureBeat, June 2026.

Alibaba has emerged as the most prominent new entrant. HappyHorse 1.1, released June 22, 2026, per VentureBeat, holds the #2 position on the Artificial Analysis Video Arena with an ELO of 1,444, leading Veo 3.1 by 69 points in text-to-video. HappyHorse supports image-to-video and multi-image reference generation through a 15-billion-parameter unified Transformer. However, the Pentagon added Alibaba to its list of Chinese military companies on June 8, 2026, per VentureBeat, adding procurement complexity for Western enterprise buyers. Google DeepMind's own benchmarks, last updated October 2025, show Veo 3.1 leading on overall preference and visual quality in head-to-head human evaluations on the MovieGenBench and VBench I2V benchmarks, though no independent update to those comparisons has been published since HappyHorse's arrival.