The mixture-of-experts trick changed how people think about LLMs. Instead of running every parameter on every token, you activate a small fraction of the network per forward pass and somehow the quality stays competitive while the compute drops. It's the reason models like Mixtral punched above their weight. Everyone in the LLM space understood it immediately. Nobody had done it openly for image generation. Until now. Nucleus-Image is a 17B parameter diffusion transformer that activates roughly 2B parameters per forward pass. It beats Imagen4 on OneIG-Bench, sits at number one on DPG-Bench overall, and matches Qwen-Image on GenEval. It's also a base model. No fine-tuning, reinforcement learning or human preference tuning. What you're seeing in those benchmarks is raw pre-training performance. That's either impressive or a caveat depending on what you need it for, probably both.

There are genuinely impressive open source music generation models out there right now. ACE Step, YuE, HeartMuLa, models that generate full songs with vocals, structure and emotion. If you want a complete track from a single prompt those are worth exploring. Foundation-1 does not compete with them. It does not try to. What it does instead is something more specific and honestly more useful for anyone who actually makes music. It generates individual loops and samples like tempo-synced, key-locked, bar-aware, built to drop straight into a production without fixing anything first. Just clean, structured instrumental loops that behave like something a producer built rather than something an AI guessed at. If you have ever spent twenty minutes trying to make an AI-generated loop fit your track you already understand why that matters.

Voxtral TTS supports nine languages: English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi, and Arabic. That by itself isn’t unusual anymore. A lot of models claim multilingual support. The interesting part is how it handles switching between them. Mistral says it can move between languages mid-sentence without changing the speaker’s voice. So you don’t get that awkward reset where the tone or identity shifts when the language changes. If that holds up, it’s actually useful for real scenarios like think support calls where people naturally switch languages, or content that mixes languages without warning.

We've watched open source absolutely run through video generation, image generation, audio. Every few months another closed model gets matched, then beaten, by something free on GitHub. But world generation always felt different. Like that was the one thing that needed a Google-sized lab behind it. I thought so too, until I actually went looking. Turns out there are open source models right now that take a text prompt and build you an explorable, interactive world. Some go even further — hand them a single image and they'll construct an entire environment around it. The quality on a few of these genuinely caught me off guard.

Text rendering in open source AI image generation has been broken for a long time. Ask most models to put readable words on a poster, lay out a comic panel, or generate anything where the text actually has to make sense and only few models can do it accurately and from rest you get something that looks like it was written by someone who learned the alphabet from a fever dream. ERNIE-Image is Baidu's answer to that specific problem. It's an 8B open weight text-to-image model built on a Diffusion Transformer and it's genuinely good at dense text, structured layouts, posters, infographics and multi-panel compositions. It can run on a 24GB consumer GPU, it's on Hugging Face right now, and it comes in two versions, a full quality model and a turbo variant that gets there in 8 steps instead of 50.

OpenAI dropped their latest model yesterday and buried inside the benchmarks is a number that deserves more attention than it's getting. On GDPval, a test that puts AI agents through real professional tasks across 44 actual occupations, GPT-5.4 matched or outperformed human professionals 83% of the time. The previous version sat at 71%. That's not a small jump. And this isn't GPT writing emails or summarizing documents anymore. This version can move a mouse, click buttons, fill out forms, and work across applications the way a person sitting at a desk would. It scored 75% on OSWorld, a benchmark that tests exactly that. The average office worker scores 72.4%. The model is already better at operating a computer than most people who use one for a living & 83% is just the beginning of what this release actually means.

If you have been looking for open source tools to work with video using AI you have probably noticed something. Most of what gets covered is generation like creating new videos from scratch. The editing side, actually modifying existing footage with AI, has been much quieter. That is starting to change. There are now open source models that can swap outfits, replace backgrounds, remove objects, change characters and apply styles to existing video using plain text instructions. Some are built specifically for editing. Others are generation models that fit naturally into a creative video workflow. This list covers both honestly. Three models built specifically for video editing and two generation models worth knowing about if you are working with video content. All open source, all available today.