New open-source and open-weight models like Llama, Qwen and DeepSeek, plus datasets, frameworks and developer tools worth trying.
Researchers just open-sourced InfiniSplat v2, a feed-forward 3D reconstruction model that turns multi-view photos or long driving videos into fully explorable 3D scenes. By decoupling geometry from appearance prediction, it cuts rendering errors and suppresses noisy artifacts. The entire model weighs around 6.6GB and runs locally on consumer GPUs.
Fish Audio just dropped Drama 3, an open-source 35B MoE speech model built for complex audio storytelling. It can juggle up to 16 distinct character voices in a single script, add contextual tags like sighs or laughs, and patch incorrect words without re-generating the entire audio track. It's basically a full cast in a single model weights file.
Researchers from Tencent and HKU just open-sourced Multix, a world model capable of simulating multiple players interacting in the same environment. It generates individual camera views while syncing player states, cutting computation by 86% with hierarchical communication. Trained on just 4 players, it scales up to 10 without extra training.
Developer Brandon Thomas used Anthropic’s Opus 5.5 to reverse engineer Microsoft Office, releasing open-source desktop clones for Word, Excel, and PowerPoint on macOS, Windows, and Linux. The core applications are free, though ArtCraft charges a subscription fee for built-in generative AI features.
Skywork just open-sourced Matrix Game 4.0, an AI world model that generates playable 3D games in real time. Instead of hallucinating everything from scratch, it offloads physics to a traditional game engine and uses AI to rerender low-poly frames at up to 83 fps on a single consumer GPU, complete with an agent that edits gameplay code on the fly. Why it matters: it solves the visual warping that plagued earlier AI games, proving hybrid neural rendering actually works.
Open-source project d2r-mobile just dropped a compatibility layer that lets users run their own Nintendo Switch copy of Diablo II: Resurrected directly on iPhone, iPad, Mac, and Android. It bypasses emulation to execute native code at a locked 60 fps with rebuilt touch controls under an MIT license.
Nvidia just open-sourced Stream Render H3, a real-time diffusion system built on MiniMax H3 that replaces traditional game rendering. The engine only tracks physics and outputs flat semantic maps, while the AI generates textures, lighting, and customizable visual styles. Catch is, running it live takes 8 enterprise GP200 GPUs and over 200GB of storage.
Researchers just open-sourced UniForge 3D, a framework that generates complete 3D assets from just a handful of unposed photos. By pairing visible-surface reconstruction with generative shape completion in a shared anchor space, it fills in unseen angles while keeping fine details sharp.
Researchers just open-sourced Pure, a 3D reconstruction model that transforms ordinary video footage into interactive dynamic 3D environments. Instead of relying on expensive 3D annotations, it post-trains on regular video by extracting monocular depth and motion cues. Clocking in at just 13.2GB with training scripts included, it makes dynamic 3D mapping practical on local machines.
Google Research just open-sourced Harness², a framework that recursively tunes eight agent harness components—such as prompts, tools, and subagents—without altering model weights. In benchmark tests, a single step improved models like Gemini 3.5 Flash and GLM-5 by up to 11.5 points.
REA just launched an MCP server that lets agents like Claude Code and Cursor reverse engineer everything from native binaries to Electron apps. All analysis runs locally on your machine, backed by real assembly, call chains, and strings instead of AI hallucinations.
CopilotKit just dropped Open Intelligent UI, an open-source alternative to ChatGPT's dynamic interface builder. It lets any agent generate charts, sandboxed widgets, or native components directly inside your app while keeping your existing design system. You're no longer locked into OpenAI to get generative interfaces.
Meta just introduced GenIA, a framework that turns standard photos or video clips into complete 3D objects. Built atop SAM 3D, it continuously checks generated shapes against source images to lock down accurate appearance, position, and scale. Why it matters: It bridges generative 3D models with real-world fidelity without retraining the base model, and the code is open for non-commercial local use.
A solo developer just open-sourced Tonefold, a free desktop composer for macOS and Windows that turns text prompts into multi-track music. Instead of generating flat audio files, it writes 11 instrument layers directly into an editable piano roll using Claude or local Ollama models. Why it matters: musicians get full control over every note, velocity, and MIDI export without dealing with black-box audio generators.
A developer just released seven open-source alternatives to Adobe's flagship apps, using Anthropic's Claude Opus 5.5 to write clean-room clones in Rust. Dubbed Artcraft, the suite mimics Photoshop, Premiere, and Lightroom with zero Adobe code under MIT/Apache licenses. While complex PSD files and plugins are still half-baked, it shows how AI is dismantling the moat around expensive legacy software.
Liquid AI just released d1-3B, an open-source multimodal decision model built on its LFM2.5 architecture. Instead of generating conversational text, it evaluates states and questions in a single forward pass with zero output tokens, clocking in at 8 ms on an RTX 4090 and 50 ms on a Jetson Orin Nano. It provides lightning-fast triage and agent guardrails directly on edge hardware.
Musubi just released PolicyLM-1.7B, an open-weight content moderation classifier under the Apache 2.0 license. Instead of relying on static categories, it evaluates your custom policies on every message in under 100ms (35ms median on an L4 GPU). It provides instant, retraining-free guardrails for live chat and AI apps.
Google just launched EmbeddingGemma 2, an open-source 740M multimodal embedding model built on Gemma 4 under the Apache 2.0 license. It maps text, code, images, video, and audio into a unified space, requiring as little as 567MB RAM on mobile hardware. Why it matters: Developers can now run privacy-first, fully offline multimodal RAG pipelines on-device without cloud calls.
Researchers just open-sourced MotionSpaceFlow (MSFlow), a lightweight model that turns text prompts into realistic human motions. Weighing around 1.1GB, it runs on consumer GPUs and lets creators pin any joint to a specific coordinate at any frame without task-specific training.
Reflection AI just introduced Beam, an open-weight 501B MoE model with 23B active parameters tailored for coding and agentic workflows. Trained across 10.5K Nvidia GB300 GPUs, it matches frontier open models like GLM 5.2 on key benchmarks while using 3 to 4 times less inference compute. Weights drop later this month.
