I've been tracking their commits and honestly, Google is notoriously slow when it involves their safety filters and compute scaling. If you need the resolution promised in 4.0 for a mid-December deadline, you're cutting it way too thin. They usually wait for I/O or a dedicated AI event, and there is nothing on the calendar before your project starts. VEO's main holdup is likely the temporal consistency at 4K which still looks a bit trippy in the private betas. If you've got $1500 to play with, you should definitely branch out now. Here are a couple of tools that actually work for high-end commercial pipelines:

• Runway Gen-3 Alpha Turbo Plan is probably the industry standard for consistency right now.
• Kling AI 1.5 High Quality Tier handles human movement with way less morphing than what you see in the current VideoFX build.
• Pika 1.5 Professional Plan is decent if you need specific physics-based shots for products. Dont get stuck in the ecosystem trap... better to have a finished project on another platform than a half-baked one on Google. Let me know if you need help with the API side for any of those.

Maybe in this Summer 2026.

Honestly, nobody really knows the exact date for VEO 3.5 or 4 yet. Google is being super quiet like they usually are before a big shift. I would suggest not waiting around if your project is in December... thats cutting it way too close for comfort. It sucks because the Gemini integration is so smooth, but the timeline is just a guessing game right now. Ive spent a lot of time testing these lately and you might want to consider these backups:

• Runway Gen-3 Alpha High Fidelity is my current go-to. The temporal consistency is solid, but be careful with the pricing tiers because those high-res exports eat credits fast.
• Luma AI Dream Machine Pro is another one Ive had success with. Its great for realism, but make sure to double check the physics in your previews because it gets a bit wonky on complex shots. I would personally pivot now just to be safe. If you need any help with the workflow transitions let me know!

To add to the point above: honestly, Id be a bit cautious about jumping ship to another big brand right away just because Googles dragging their feet. It might feel like the grass is greener with the other major players, but those subscription costs add up fast and you might end up paying for pro features you dont even need yet. Most brands in this space have totally different pricing structures and safety filters that might just annoy you more than the wait. Before you make a move, I had a couple questions to help narrow it down:

• What exactly are you trying to generate? Like, are we talking hyper-realistic human movements or just some stylized backgrounds?
• Whats your actual budget look like for this project? I would suggest really thinking about whether you actually need the 4.0 specs or if a cheaper, existing tool could handle the heavy lifting while we wait for Google to get their act together. Sometimes these companies promise the moon and deliver a pebble... just something to keep in mind.

> Honestly, nobody really knows the exact date for... I get the frustration but honestly, in my experience, waiting for a release date from Google is a losing game. Ive spent years watching these cycles and they almost always prioritize safety tweaks over actual usability when they finally do drop something. You really shouldnt keep your project on hold for tech that doesnt exist yet. My advice? Just go with Runway. You can pretty much pick any of their high-end options and get the reliability you need right now. I dont think its worth the stress of tracking commits when you could be getting work done with a platform that actually has its stuff together. Just get a subscription with Runway and call it a day, honestly.

Regarding what #1 said about "Honestly, nobody really knows the exact date for..." - they're spot on. I've been tinkering with local video generation pipelines for a while now, trying to build my own solution because the wait for VEO is just painful. Last month I spent way too many hours trying to optimize some open-source weights on a custom workstation I built, and honestly, the bottleneck is always the VRAM when you try to scale up the resolution. I think I heard somewhere that Google is hitting a similar wall with their server clusters for the 4.0 version. Not sure but a buddy of mine who follows the dev commits closely mentioned they're reworking the entire sampler logic right now. IIRC, the last time they did a major overhaul like this, the release got pushed back by months with zero warning. It's kinda the DIY way or the highway right now if you actually want to get work done. Waiting for a corporate release date is basically just gambling with your productivity. Just my two cents from someone who usually prefers to just hack something together rather than waiting for a polished UI that might never show up.