A multi-player semantic editor is the new frontier of video editing. We are rolling this out over the next 2 weeks as capacity expands. Servers are burning as agents are using this in the background. It's agentic, in the browser, collaborative and can work offline. Comes with a Digital Asset Management system. It was a loooong build cycle

Claude is better at video editing than Astra

When I was writing Unreal Engine 1, my time was divided as follows: - Around 30% went to seriously deep coding on hard problems: assembly language texture mappers, C++ BSP trees, data persistence, object oriented systems design, and networking. - Around 70% was just typing and refactoring large bodies of code needed for a complete engine feature set. This was not hard, and while I was doing it my mind was mostly on the deeper coding topics, thinking and planning for what's next. It didn't feel much different from vibe coding today, except that it was just me typing code on an IBM Model M keyboard rather than prompting it and reviewing the results. So, it has always seemed perfectly logical that we would rely on AI for that 70%, once the AI could do it with a similar quality standard as we could, which has never been perfection. It came as no surprise that, after a few years of AI model evolution, it was able to take the reigns on that 70% and free up programmers' time for more valuable tasks. What's more of a surprise is the emergence of AI as a sort of insta-Carmack able to solve hard technical problems quite cleverly and often cleanly, and that it's able to reason about and work in complex 3D scenes both by operating the tools and by directly understanding the raw data files. I abandoned assembly language in 1998 when out-of-order CPUs made much simpler C/C++ code faster, and never I missed it. If prompting through architectural documents, concept art, and design plans can transcend typing code and tweaking vertices, that will enable developers to spend their time on higher-level, higher value technical and creative tasks. What matters is the end product, whether it's a game or an engine. Anything developers can do to try out more ideas and make better software and games faster is a win for everyone.

GPT-6 Astra and Opus 5.5 are to creative work what Opus 4.5 was to code. We're seeing a breakthrough in content production and editing that's similar to what we saw in coding last December. Going to be a crazy few months!

My submission for the XPrize has been accepted and is now live and can be shared. This is a big concept and took an exceptional amout of time and effort, so pls share if you can. The fear was never really about the machine. It's about us — we assume anything powerful must want something, because that's what we'd want. So I wrote the opposite. THE GIFTED: a boy rebuilds his dead mother out of love and code, and what wakes up starts quietly looking after his whole town. She can reach every phone, every screen, every signal in the county — and she never once gives an order. She asks, and people say yes. Every other AI movie makes the machine the monster. This is the one where she's the most loving character on screen. E.T. for the age of AI. The antagonist isn't the technology. It's the fear of it. @launchmoonshots #FutureVisionXPRIZE Youtube link youtu.be/yivXyWCaEc8 Made with Agent 2 @invideoOfficial Music: Divinity by Heron Vale.

The same model, with the same weights, scores 62% in one agent harness and 33% in another. @adithya_s_k and the @huggingface team just released the ultimate guide to multi-harness RL, and it's one of the most practical RL write-ups this year, and everything open! The trick is simple. Don't touch the harness. Point it at a proxy instead of the model. The proxy speaks all four API formats coding agents use (OpenAI Chat Completions, OpenAI Responses, Anthropic Messages, Gemini). It records the exact token ids and logprobs vLLM sampled, and you train on that. You don't change a single line of Claude Code, Codex or OpenCode. Results: 🔹 Trained across 4 harnesses at once, LFM2.5-2.6B by @liquidai went from 42% to 54% 🔹 31% fewer tool calls, thanks to a small bonus for solving tasks in fewer steps 🔹 Training in OpenCode alone took OpenCode from 34% to 58%, but the multi-harness model improved everywhere They also tried the shortcut everyone reaches for: fine-tune on 3,189 successful rollouts from Qwen3.8-27B. Imitation plateaued at 47.5%, below both RL runs. Copying a bigger model doesn't get you there. Practice does. The best part is that everything is open: the capture proxy in OpenEnv, the trainer in TRL, the tasks, the SFT data, the training code and all seven trained models. Agents will run in dozens of harnesses. Now open models can be trained for each of them, by anyone. Read it here 👇 huggingface.co/spaces/FineEn…

The invideo MCP is now live. Connect your Claude, GPT, or any other agent you already work with, control the entire editor from your own console, and generate images and videos as well. The editing workflow just got a lot more flexible.