Veo 2 Rolls Out to Gemini, Kling 2.0 Stuns with Global Launch, and Open Source Video AI Pushes the Boundaries

COEY’s Weekly AI Video Generation Report

This week the generative AI video space continues its high-velocity evolution, driven by breakthroughs from U.S. and Chinese giants and a surge of new open-source models. Below are the key stories shaping the landscape, each backed by releases and news from the past seven days.

Google Veo 2: Now in the Hands of Millions Through Gemini

Google’s Veo 2 text-to-video generator has completed its global rollout to Gemini Advanced and Google One AI Premium users. The news cycle saw an explosion of hands-on coverage as Advanced subscribers around the world gained access to 8-second, cinematic-quality, 1080p video creation directly from Gemini’s interface. All the latest reviews agree: Veo 2 marks a massive leap in accessibility and user experience. The suite includes:

• Outpainting and inpainting for targeted region edits and scene extension

• Templates for travel, social campaigns, and mini-documentaries

• Rapid, intuitive text-to-video translation at $0.50 per second

What sets this moment apart is the everyday creator’s ability to generate pro-quality video securely and swiftly, all from the Gemini app. The learning curve is minimal; results are consistently cinematic for a wide array of prompts. This mainstream debut has provoked a wave of creative projects in business, marketing, and the arts, showing Google’s pivot from headline-grabber to staple creative tool. Details here.

Kling 2.0: Kuaishou’s Global Challenge Lifts the Bar

China’s Kuaishou has propelled Kling 2.0 to the global stage, the most powerful and accessible text-to-video platform to date. The past week saw Kling 2.0’s user base soar, exceeding 22 million, and its creative challenge campaign go live to encourage user-submitted films and scenes. Recent releases confirm:

• Hollywood-caliber photorealism, lighting, and motion

• Consistent character and background fidelity across extended sequences

• Advanced multimodal workflows mixing text prompts, drawn inputs, and images for nuanced editing

• Extensive international access, with new tools driving rapid growth in creative participation

This democratization of AI-powered filmmaking enables creators worldwide to experiment with sophisticated editing, previously the sole province of elite studios. Kling's features now directly compare to Sora and Veo 2, with many scenes revealing parity or superiority in visual realism. Read more.

Open Source Video Models: Moonshot AI’s Kimi-VL Lands

Open-source innovation has stepped firmly into the limelight. The headline this week is Moonshot AI’s release of the Kimi-VL model, a 2.8-billion parameter vision-language system designed to handle text, images, and video with unprecedented affordability and efficiency. In just days, it’s sparked adoption among developers aiming for scalable, nimble multimedia pipelines, especially in resource-constrained environments. Notable features:

• Supports multimodal reasoning (text-image-video)

• Runs on consumer GPUs with rapid inference times

• Open weights and transparent licensing fueling rapid remixing across Asia and beyond

This lightweight model is quickly becoming a global favorite for academic research and start-up video projects, a key step in the arms race for the world’s best open-source video solutions. More here.

Alibaba’s Wan 2.1 Remains a Benchmark, No New Releases This Week

Despite ongoing buzz, no major new version of Alibaba’s Wan 2.1 video generation suite landed in the past seven days. However, the open-source suite, released in February, continues to trend on usage charts. Latest global benchmarks and user stories report its outputs rival closed, proprietary models for certain genres, testament to the gap-closing power of open-source AI, particularly by Chinese teams.

Runway Gen-4: Scene Consistency and Reference Tools

On the U.S. front, Runway’s Gen-4 has not debuted a significant new model or major features update this week. Yet recent coverage continues to highlight its industry-defining work in scene and character consistency. Production teams rely on Gen-4 for:

• Reference-based tracking of objects and style across sequences

• Turbo rapid-prototyping tools for instant video iteration

These advances keep Runway at the top of professional content creation, holding strong amid fierce competition, though the last major upgrades were earlier in April. Source

Industry Trends and Fresh Faces

• Rapid price drops and aggressive licensing (in both open and closed systems) continue to fuel mass adoption, especially in regions under-served by U.S. models.

• Meta’s MoCha AI tool (April launch, still trending) lets users animate characters via text and voice, highlighting the vertical surge in AI avatar creation.

• Most recently, open-source and multinational collaborations in Asia are hastening cycles of model remixing, benchmarking, and deployment, jolting the pace of global progress.

The Takeaway: The Global Quality Gap Narrows

The clearest outcome of this week’s news is the closing gap in quality, price, and accessibility between closed and open AI video systems. With Veo 2 and Kling 2.0 in broad user hands and new models like Kimi-VL supercharging open science and indie development, the creative future grows bigger, and more unpredictable, by the week. Watch this space for what’s next.