Gemini Omni Leaks, Kling AI's $20B Spinoff, and the New Benchmark Battle: What's Moving the Image & Video Stack Right Now
News
A UI string reading "Powered by Omni" surfaced inside Gemini's video tab next to Veo 3.1's internal codename, pointing to a new unified model that generates text, images, and video in a single pipeline — a first for any top-tier AI product. With Google I/O just five days away, the leak also surfaced a companion codename "Spark Robin" as a possible dedicated image layer, suggesting Google is consolidating its fragmented Imagen/Veo/Nano Banana stack into one coherent story.
Kuaishou Confirms It's Evaluating Kling AI Spin-Off at $20B Valuation — Shares Surge 10%
Kuaishou's board filed a formal disclosure with the Hong Kong Stock Exchange acknowledging it's evaluating a restructuring for Kling AI that could involve external financing and an independent listing, with a Pre-IPO round targeting a $20B valuation and Tencent reportedly in discussions. Kling's ARR has reportedly hit $500M — up from $240M in December 2025 — making this one of the fastest revenue ramp-ups in the generative video space and a clear signal that AI video is becoming a standalone asset class.
Analysis
Three Ways to Read the Gemini Omni Leak — and What Each Means for Builders
WaveSpeed breaks down the three plausible interpretations of the Omni UI string: a rebrand of Veo, a new Gemini-trained video model running alongside Veo, or a true unified omni-modal architecture. The most disruptive reading — a single model handling image and video generation in one context window — would make the current multi-model API stack look unnecessarily complex and could reshape how production teams budget compute.
Kling AI's $500M ARR Trajectory and the Case for a Pure-Play AI Video IPO
The detailed ARR breakdown shows Kling going from $20M monthly revenue in December 2025 to a $500M annualized run rate by April 2026 — a roughly 2x acceleration in under five months. Morgan Stanley's endorsement of the spin-off structure matters: it validates the market thesis that pure-play AI video companies command meaningfully higher multiples than video features buried inside a social platform.
What Is Gemini Omni? A Practical Pre-I/O Evaluation of the Leaked Video Generation Quality
A demo video shared publicly on May 11 gave the first real look at Gemini Omni's generation output — reviewers noted strong video remixing, object replacement, and scene rewriting capabilities that no other AI video tool currently does well, even if raw visual fidelity still trails Seedance 2.0 and Kling V3.0. For builders, the strategic read is: if Omni's conversational editing and template-based workflows hold up at launch, it could become the default for high-volume social and ad content even without leading on cinematic quality.
Tools
Seedance 2.0 vs. Kling 3.0: Practical API Pricing and Workflow Decision Guide
fal.ai's head-to-head breaks down the real cost delta between the two current benchmark leaders: Kling V3 Standard at $0.084/sec (audio off) vs. Seedance 2.0 fast at $0.2419/sec, with Kling outputting at 1080p versus Seedance's 720p on the platform. The tactical takeaway for builders is clear: use Kling V3 Standard for rapid iteration and storyboarding, route final audio-synced, multi-reference productions through Seedance's standard tier where its unified audio-video pipeline justifies the ~3x cost premium.
Stay Ahead
Delivered each morning.