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.

Lumina-DiMOO is a state-of-the-art open source multimodal AI system, designed as a completely free and flexible Nano Banana alternative. This model is capable of text-to-image generation, image editing, inpainting, style transfer, subject-driven creation, controllable generation, extrapolation, and advanced image understanding, all in a single, developer-friendly framework.

Google's spending billions training AI models. OpenAI's hiring armies of engineers. And somehow, a startup in Bengaluru just outperformed both of them. Sarvam AI's new Vision model scored 84.3% on olmOCR-Bench—a brutal test that makes AI models read messy scanned documents, handwritten notes, and complex tables. Google Gemini 3 Pro got 80.2%. ChatGPT? A distant 69.8%. If you're thinking "okay, cool benchmark, but who cares?"—fair. Here's why this matters: billions of documents across India are locked away in regional languages. Government records in Gujarati. Medical files in Tamil. Historical archives in Bengali. The big AI models can read these languages, but they mess up constantly—wrong characters, mangled words, useless output. Sarvam doesn't. It's specifically trained on Indian scripts, and the results show. For the first time, Indian companies have an AI tool that can reliably digitize documents in 22 languages without sending everything to Google or OpenAI's servers.

Most people assume beating a Google model requires another massive frontier model. More parameters. More compute. That is just how the hierarchy usually works. Reka Edge is a 7-billion-parameter model. Yet it manages to outperform Gemini 3 Pro on object detection benchmarks, and with quantization it can even run on devices like the Samsung S25. That combination should not exist. A model small enough to fit on a phone outperforming a frontier AI system from Google on a specific but genuinely useful task is not something you expect to see in 2026. Yet here we are. This is not a model that beats Gemini at everything. It does not. But where it wins it wins convincingly.

Most small AI models are a compromise. You give up reasoning for size, or vision for speed. Qwen3.5-4B doesn't seem to have gotten that memo. Alibaba just dropped Qwen3.5, and the 4B version is the one worth paying attention to. It thinks before it answers, reads images and video, handles 201 languages, and sits on a context window of 262,144 tokens, longer than most models ten times its size. All of that in something small enough to run on your own machine. I went through the benchmarks and architecture docs so you don't have to. Here's what actually matters.

The company that sells the chips just built software that runs on everyone else's chips. Nvidia is reportedly preparing to launch an open source AI agent platform called NemoClaw at GTC 2026 next week in San Jose. People familiar with the plans say the platform will let enterprise companies deploy AI agents across their workforces regardless of whether they run on Nvidia hardware or not. Nvidia hasn't confirmed anything publicly yet. But the conversations with companies like Salesforce, Cisco, Google, Adobe and CrowdStrike are apparently already happening.

Most open source models that claim agentic capability are really just instruction-tuned models with tool calling bolted on. They can call a function. They cannot think across ten steps, remember what they decided three tool calls ago, and course correct when something breaks mid-task. This is where Trinity-Large-Thinking comes into picture. Arcee AI released it this week. 398 billion total parameters, but only 13 billion active during inference. That MoE architecture means it runs closer to a 13B model in practice while carrying the knowledge of something nearly 30 times larger. And unlike most models where reasoning stops between steps, Trinity keeps its thinking tokens alive across the entire agent loop. Every decision it makes is informed by everything it reasoned through before it.