Introducing MiniMax H3 by MiniMax Design (H3) Our next-gen... open-weight multimodal video model, built for general intelligence beyond single-task generation. What makes it different: -Native multimodal understanding & generation -Precision editing & control -Commercial-grade output for film, ads, MVs, UI, game CG -Cinematic quality: native stereo audio, up to 2K/24FPS -Multi-asset reference + voice cloning = idea to finished product, in one flow. Try H3 now → #MiniMaxH3show more

MiniMax H3 (Hailuo-03) is now available on fal. -... Open-weight multimodal video model with native stereo audio on every generation. - Combines up to 9 images, 3 video clips, and 3 audio clips as references - Holds subjects, motion, and audio consistent - Weights will be released soon.show more

MiniMax H3 is on Leonardo. Most video models give... you a clip. H3 gives you the clip and the soundtrack — with your character and voice locked in from refs. Expect: - Commercial-grade quality across ads, e-commerce, gaming & UI - Best value in its class - One lightweight model, fully multimodal Built for one-shot finished content: brand teasers, product videos, fashion films, talking charactersshow more

MiniMax H3 is a serious upgrade for AI video... creation. Instead of relying on just text prompts, you can combine images, videos, and audio references to control the character, motion, camera, and sound. What stands out: → Up to 9 image + 3 video + 3 audio references → Native synced voice, music & sound effects → 2K video generation up to 15 seconds → Instruction-based video editing → Motion transfer → Better character, camera & voice control The interesting part is that you can build a full scene from references instead of endlessly regenerating until the result feels right. 🔗 Now is a great time to try MiniMax H3 on Magnific, especially with the current 2K offer.show more

MiniMax H3 is now 50% OFF on Magnific for... 2K video, only until September 1. 🔥 I’ve been trying MiniMax H3 on Magnific, and it feels like a big upgrade for AI video creation. It’s not just about turning text into videos. You can use text, images, videos, and audio together in one prompt, giving you more control over the final video. Here’s what makes it stand out: - Multimodal: Use text, images, video, and audio in one prompt. - Multiple references: Add up to 9 images, 3 videos, and 3 audio files. - 2K video: Create videos up to 15 seconds long. - Built-in sound: Generate voice, music, and sound effects with the video. - Easy editing: Remove objects or transfer motion easily. - More control: Control the camera, characters, and voice. You can use it to turn posters into videos, moodboards into short films, and product images into ads. It also helps bring your ideas to life with realistic movement, lighting, reflections, and sound. The workflow is simple: give it your references → generate → edit → refine. Try MiniMax H3 on Magnific:show more

The next generation of AI video has arrived. Seedance... 2.5 combines long-form video generation, native audio, and precise scene control in a single model. — Text-to-video & image-to-video — Up to 30-second videos — Native synchronized audio — First & last frame guidance — Built for cinematic content Try now →show more

MiniMax H3 is now on Magnific and honestly, there’s... a lot you can do with it. You can mix text, images, videos and audio in a single prompt up to 9 images, 3 videos and 3 audio references. Start creating now: It can generate up to 15s of 2K video with synced sound, including voice, music and effects. And you can go beyond generation too: edit clips, remove objects, transfer motion, and control the camera, character and voice. But the multi-reference workflow is probably my favorite. Give it your product, character, environment and motion references, and H3 pulls everything together. It feels like a much easier way to go from an idea to an actual finished video.show more

Seedance 2.0 is coming to Elser AI 🎬 Early... access → This isn’t just an update. It’s a studio-grade AI film engine upgrade. • Multi-shot storytelling • Long-form character consistency • Native audio-visual generation • Multimodal input + in-model editing • Advanced motion synthesis Built for real cinematic AI filmmaking. Big shift 👇 UNLIMITED 2-Year Creator Plan — 87% OFF One subscription. Seedance 2.0 + Kling 3 + Nano Banana Pro + Veo 3.1. If you’re serious about AI anime & cinematic storytelling — this changes the game. Early access →show more

I tried MiniMax Design (H3) to see how it... handles real content creation. The workflow is simple. You just write a prompt or drop in an image, and it turns that into a dynamic video with motion, framing, and scene depth. No timeline to manage. No editing setup. No back and forth. What stood out to me: • Text to video and image to video both feel smooth. • It handles motion, camera angles, and flow on its own. • Output is fast, usually within seconds. • Works well for reels, quick ads, storytelling, and idea testing. It removes the hardest part: starting from scratch and turns your ideas into content in minutes. Instead of thinking, “How do I make this video?” You start with, “What do I want to create?” That shift alone makes it worth exploring. Try it here: #Hailuoshow more

llama.cpp isn't just for text LLMs anymore. Pure C++... zero shot voice cloning just officially landed in mainline. Text generation was only step one. If you’re building autonomous local AI agents, real time voice assistants, or edge workflows, instant low latency audio is the missing piece. Thanks to PR #26254, Alibaba’s state of the art Qwen3 TTS model family is now natively supported directly inside the llama.cpp repository under the multimodal (mtmd) framework. No Python bloat. No massive PyTorch CUDA overhead. Just raw, hyper optimized C++ running GGUF voice weights. Here is why this native update is a massive deal for the open source local AI stack: # Multimodal Architecture (.gguf + mmproj) Qwen3-TTS splits the workload between the base language model backbone and a multimodal projection adapter. llama.cpp handles this using the llama-tts binary, mapping the text model alongside its --mmproj projector to process audio tokens seamlessly. # Zero Shot Voice Cloning in Seconds You don't need fine tuning or massive dataset training. Feed the C++ engine a single 5 to 10 second .wav audio sample using the --tts-speaker-file flag, and it accurately clones the exact timbre, tone, and accent on the fly. # Real World T4 GPU Benchmark & Resource FootprintRunning the 1.7B Base model in 8-bit quantization (Q8_0): - VRAM Footprint: ~7 GB peak VRAM during active zero-shot cloning. - Audio Quality: Studio grade, natural-sounding voice output in seconds. • - Execution: Direct execution via native compiled binaries or sub process calls. # Coming Next to llama-server (PR #26603) Beyond CLI execution, a native POST /tts HTTP endpoint is currently being added to llama-server, which will soon allow you to trigger voice generation directly via standard REST API requests! # quick note on Colab compilation: Because this code was merged into mainline very recently, pre-built third-party binaries haven't fully caught up yet. Compiling llama-tts directly from source on Google Colab's free CPU instance can take about 1 hour (or ~1-2 minutes if targeting single GPU arch like -DCMAKE_CUDA_ARCHITECTURES=75). Be patient during the build step, or compile it locally on your own rig for instant execution! To test this out yourself, I built a zero config Google Colab notebook that compiles llama.cpp, downloads the Q8_0 GGUF files from HuggingFace, and spins up an interactive Gradio Studio UI so you can record/upload 3 second clips and clone voices in real time. Stop sleeping on native C++ audio. The era of bulky Python audio pipelines is officially over. Links to the free Google Colab notebook and the official ggml org GGUF HuggingFace model repository are in the replies below! available in q4 and q8 both variants, 1 GB and 1.85 GBs respectively (requires additional ~500MB mmproj gguf) Are you building local voice agents yet? What does your current audio stack look like? Drop your setups below!show more

I've been testing the new Seedance 2.5 model, and... it's now available on CapCut. What I like most is that I can generate and edit AI videos in one place instead of switching between different tools. The workflow feels much more seamless, especially when refining longer projects. It also gives creators: → More controllable video generation & editing → Timestamp-based storyline control → Support for up to 50 references → Up to 90-second video generation (30s native + 2 extensions) → Improved multilingual performance Experience Seedance 2.5 on CapCut today: → Try it on the web: → Get the app: #CapCut #Seedance25 #CapCutai #CapCutDidThatshow more

Big moment for text-to-speech. Qwen just open-sourced a text-to-speech... model that lets you clone voices, design new ones, and control speech using natural language. Let me explain what I mean: You can literally tell it "speak in a cheerful tone with slight nervousness," and it actually does that. No complex audio engineering needed. What makes this special: - 3-second voice cloning - Covers 10 languages: English, German, French, and more - Latency as low as 97ms for real-time applications - Supports both streaming and non-streaming generation The model comes in two sizes (0.6B and 1.7B parameters), so you can pick based on your hardware and quality needs. Three modes to work with: 1. Custom Voice: Use pre-built premium voices with instruction-based style control 2. Voice Design: Describe the voice you want in plain English (or Chinese), and the model creates it 3. Voice Clone: Provide a 3-second reference audio and clone that voice The best part? It integrates with vLLM for production deployment and has a simple Python package you can pip install. I've shared a link to the GitHub repo in the next tweet.show more

We are in an insane run of open-weight drops.... Every modality, open source is winning. This is what an open source AI summer ☀️ looks like: 🧠 LLMs & Reasoning → DeepSeek-V4-Flash-0731 (my king 👑): 304B MoE refresh, Terminal-Bench 2.1 jumps 61.8→82.7 over the preview, DeepSWE 7.3→54.4. Closes in on Opus-4.8 on Agents' Last Exam (25.2 vs 25.7). MIT. → Muse-Glimmer-30B, from Meta (they are back!!): their first open agentic model. ~29.6B dense + perception encoder, 131k+ context, built to run fully local, no cloud. Apache 2.0. → Liquid AI LFM2.5-2.6B: 2.69B params, 131k context, 220 tok/s on an M5 Max in under 2.5GB RAM. Competitive with models 4x larger on agentic tasks. → inclusionAI Ling-3.0-flash: 124B total, only 5.1B active, ~12% the size of their old 1T flagship Ring-2.6, matches it on key benchmarks. MIT. → inclusionAI Ling-3.0-tiny: 7.9B total, 1.3B active, 86-90 tok/s on an M4 Pro MacBook at ~8GB peak memory. MIT. → NVIDIA Nemotron-3.5-Lightning-30B-A3B: hybrid Mamba-2+MoE+Attention, up to 1M context, runs on a single H100 or DGX Spark, SWE-bench Verified 52.8. → deepgrove maple-preview: 20B-A1B ternary-weight reasoner, 218 tok/s on a Mac mini M4, 5.3GB checkpoint. MIT. → BigBang-v1 (endless-frontier): fine-tuned from Qwen3.6-35B-A3B via a self-evolving generator/critic synthetic-data loop. Lands aggregate performance between DeepSeek V4 Flash (284B) and V4 Pro (1.6T), at 35B. Apache 2.0. 🎬 Video → MiniMax-H3: 33B dense omni model, native stereo audio, up to 2K/15s. 3.6k+ likes already. → Minimax-H3-Turbo (lightx2v): Apache-2.0 turbo distillation of H3 for fast inference. → Lightricks LTX-2.5: image-to-video update, custom Gemma-4-12B text encoder, a markedly stronger distilled model. 🔊 Voice → NVIDIA NemotronLabs VoiceChat-11B: full-duplex speech-to-speech, ~450ms turn-taking, #2 on open VoiceBench, and the first open full-duplex model with live tool-calling mid-conversation. 🛡️ Safety → Mistral Shieldstral-1.0-3B: 3B multimodal guardrail that takes your safety policy as plain text instead of fixed categories. Beats LlamaGuard-4-12B and ShieldGemma-9B on HarmBench (99.4) and ToxicChat (84.1) at a fraction of the size. Apache 2.0.show more

I just ran Gemma 4 31B on @CerebrasSystems at... 1,800+ tokens/sec and it's multimodal. For context: that's 35x faster than a typical GPU endpoint, and the first token (reasoning included) lands in 1.5 seconds. This isn't a benchmark slide, I recorded the inference live. Prompt I used: "Create a simulation of an iPhone. Include at least one working dummy note taking app, a functional notification pulldown, high quality graphics, single HTML file, any libs via CDN." - Generation time: 3 seconds. - Notes app worked. - Notification panel worked. - Rendered first try. This is what wafer-scale inference unlocks, not just "faster," but a different category of product. When generation is this fast, you stop waiting and start iterating in real time. Why this matters: Gemma 4 31B is Google DeepMind's flagship open weight model, Apache 2.0 licensed, dense (not MoE), and built for efficiency over raw parameter count. It scores close to Claude Haiku 4.5 on the Artificial Analysis Intelligence Index (30 vs 29) but runs ~18x faster on Cerebras. It's also the first multimodal model on Cerebras's platform, meaning you can now feed it screenshots, documents, charts, and UI states at wafer scale speed. # Applications I'm most excited about: - Screenshot → Insight: Drop in a dashboard or document screenshot, get structured findings back instantly. no waiting, no batching. - Live UI generation: Full interactive interfaces (like my iPhone sim) generated and rendered in under 2 seconds. - Screenshot -> Patch: Feed it a broken UI + console error, get a minimal code fix and verification steps back. - Computer use & agentic loops: See -> reason -> act - verify, fast enough to keep a human in the loop instead of waiting on the model. - Long context summarization: Full research reports condensed into decision ready summaries you can read and requery in one sitting. The bigger unlock isn't the speed number itself, it's that agentic and multimodal loops (see -> reason -> output -> tool call -> verify -> retry) finally run in real time instead of feeling sluggish. As Logan Kilpatrick (Logan Kilpatrick) put it: "If every model was doing 2,000 tokens per second, you wouldn't build the same product and just have it be faster, you'd build different products." Gemma 4 31B is live now on Cerebras Inference Cloud in public preview. If you're building multimodal, agentic, or real time apps, this is worth testing today. What would you build with such insane inference throughput?show more

You don't need a GPU for fast studio grade... voice cloning anymore. Qwen3 TTS (1.7B Q4_K_M) + mainline llama.cpp is officially the fastest way to generate zero shot voice clones using 100% pure CPU execution. Following up on my last post where we ran the Q8 model on a GPU, we just took local C++ voice synthesis a massive step further. The open source community quantized Alibaba's SOTA Qwen3 TTS model down to Q4_K_M GGUF, completely freeing local audio pipelines from dedicated graphics hardware. Here is the real world benchmark and hardware breakdown of running SOTA voice cloning on CPU: # Architecture & Model Setup Using Qwen3-TTS-12Hz-1.7B-Base-Q4_K_M.gguf paired with the 8 bit multimodal projector (mmproj-Q8_0.gguf), llama.cpp executes the entire pipeline in pure C++. No PyTorch, no CUDA dependencies, and no VRAM bottlenecks. # Real-World Memory Footprint - Baseline RAM: 1.6 GB system idle. - Peak Generation RAM: 8 GB RAM during active voice synthesis. - Requirement: Any basic machine with at least 8 GB of system RAM can run this easily. # Real World CPU Benchmarks - Google Colab Free Tier (Throttled 2 Core CPU): Synthesizes a 5 sec studio quality audio clip (~8 words) in 45 seconds. - Modern Consumer CPU (Intel i5/i7 13th/14th Gen or AMD Ryzen 7000/9000): generation should drop to 5 to 20 seconds (nearly 1:1 real-time generation speed!). # Zero Shot Voice Cloning Quality Pass any 5 to 20 second .wav audio sample to the C++ engine using the --tts-speaker-file flag. It yields clean, natural sounding cloned speech with virtually zero quality loss compared to unquantized FP16 weights. To make testing seamless, I built an updated zero config Google Colab notebook. It pulls the official pre built llama.cpp CPU binaries (zero compilation time!) launches a live Gradio web app right in your browser. Record a 5 second clip from your mic (or drop a .mp3, .wav file), type text, and generate cloned audio on CPU. Native C++ audio models are making edge based, offline AI voice agents a reality. Links to the free Q4 CPU Colab notebook and the Q4_K_M GGUF HuggingFace repository are in the replies below! Which models have you been running on your CPUs? What CPU hardware are you using for local inference?show more

Kling 3.0 for AI UGC videos is absolutely insane... 🤯 I spent 25,000+ credits in Kling perfecting the ultimate prompting framework to get the best AI UGC outputs possible. And the 3.0 update just made everything even better. Perfect for DTC brands and agencies who want high-quality AI UGC without paying $500/video for actual UGC. Here's what Kling 3.0 unlocks for AI UGCL → "AI Director" system that understands full scripts and auto-schedules camera angles (shot/reverse shot) in one generation → 3 to 15 second cinematic clips with full temporal coherence → Improved character and element locking so your subject stays consistent across shots and angles → Native 4K output for both video and stills — actually usable for professional ad creative And best of all: perfect character consistency ACROSS different shots. What this means for AI UGC: - Multi-shot storytelling in a single generation cycle - Longer clips that actually hold together - Consistent characters across your entire ad - Output quality that's ready for paid media This is the closest AI video has gotten to replacing a real shoot for performance creative. I recorded a full breakdown of the prompting framework I built after burning through 25,000 credits. Want access to all the prompts I use to create AI UGC with Kling? > Like this post > Comment "KLING" And I'll send it over (must be following so I can DM)show more

Wan 3.0 is live on Magnific. One generation now... takes 20 references at once: 10 images, 5 audio, 5 videos. Not one style ref. Twenty. What that actually changes: Length : 2 to 30 seconds in a single pass. No stitching, no extending clip by clip and praying the character survives the cut. Native audio : voice, ambience and SFX generated with the video, not laid on top afterwards. Multiple ratios available , and adaptive the model composes for the frame instead of cropping into it. 10 image refs : a character, a product, a location and a lighting look can all hold in the same shot. ⚡ 40% off until September 30. More details in comments 👇 What's the first thing you'd throw 20 references at?show more

This Kling 3.0 UGC workflow is absolutely insane 🤯... It generates hyper-realistic UGC-style ad videos from a single text prompt — talking heads, unboxings, testimonials, tutorials — in one shot. Perfect for DTC brands and agencies who need to scale UGC without paying $500 per UGC video. Here's the problem: You're either paying creators who take weeks to deliver, or you're stitching together janky 5-second AI clips that look obviously fake. Kling 3.0 solves it: → Write a prompt with character, scene, and dialogue → Get a 10-15 second UGC video with realistic facial expressions → Native voice control — tone, pacing, emotion per line → Multi-shot scenes in one generation (hook → demo → CTA) → Paste negative rules to kill AI artifacts instantly No creator fees. No stitching clips. No uncanny valley. What you get: → 6 copy-paste prompt templates for every UGC format →Voice & dialogue cheat sheet with tone keywords → Multi-shot ad structure (hook → problem → solution → CTA) → Image-to-video workflow for branded product shots → A "where it fails" section so you don't waste credits → ChatGPT template to convert any rough idea into a Kling prompt I built a full prompting guide for UGC ads on TikTok and Meta. Want the guide? > Like this post > Comment "KLING" And I'll send it over (must be following so I can DM)show more

Contact sheet prompting is the hottest AI video technique... right now 🤯 One image in → 6 consistent frames out → cinematic video ads in minutes. But everyone's doing it manually. I automated the entire workflow in n8n + Airtable. Here's why contact sheet prompting is blowing up: You give AI one reference image, and it generates a grid of consistent shots — same face, same outfit, different angles. Instant storyboarding, full creative control, no photoshoots. The problem? It's super tedious: → Write the prompt manually → Generate the contact sheet → Crop each frame by hand → Feed frames into a video model one by one → Repeat for every product This n8n automation handles all of it: → Upload character image + product image → AI analyzes both and writes the contact sheet prompt → Nano Banana Pro generates a 6-frame grid → System extracts each frame automatically → Kling 2.5 generates smooth transitions between frames → You get 5 video clips ready to stitch Approval checkpoints at every stage, no surprises. What lands in your Airtable: → AI-generated creative prompt → Core hero image (model + product) → 6-frame contact sheet → 5 cinematic video clips → Full control before each generation step Contact sheet prompting on autopilot. I filmed a 20 minute Loom video showing you exactly how I set it up. Want the Loom + the complete n8n workflow + Airtable base? > Comment "SHEET" > Like this post And I'll send it over (must be following so I can DM)show more

AI Is Moving Beyond “Generating Videos” — Toward “Generating... Worlds” Over the past two years, AI video models have advanced at an astonishing pace. From Runway and Pika to Sora and Veo, AI-generated videos have become increasingly realistic and more consistent with the physical laws of the real world. Many people believe the next objective is simply to generate videos that are longer, sharper, and more lifelike. But if we take a step back, we can see that the real transformation is not happening in video itself. It is happening in world models. What Is a World Model? In 1943, psychologist Kenneth Craik proposed an idea that would influence artificial intelligence research for decades. He argued that the human brain does not merely react to the outside world. Instead, it maintains an internal model of how the world works. Because we have this internal model, we can predict the outcome of an action before we actually take it. Before crossing a road, we estimate whether a car will pass by. Before catching a ball, we predict its trajectory. These abilities come from continuously simulating the world in our minds, rather than relying entirely on trial and error. This idea later became known by a more formal term: World Model. A world model does not describe a single image or a fixed video clip. It is an internal representation capable of continuously simulating the rules and dynamics of the real world. Why Is AI Research Turning Toward World Models? Because predicting “what comes next” is becoming increasingly central to how AI systems work. Language models predict the next token. Image models predict the next step in the denoising process. Video models predict the next frame. A world model, however, attempts to predict something broader: What should the world look like in the next moment? In 2018, David Ha and Jürgen Schmidhuber proposed in their paper World Models that an intelligent agent could first learn a model of the world, and then use that internal model to plan its actions. The Dreamer series later demonstrated that many complex tasks could be learned by training agents inside an “imagined world.” At the same time, the development of video models such as Sora and Veo led researchers to another realization: A model capable of continuously generating video has already learned, at least implicitly, many of the rules governing the real world. As a result, these two research directions have gradually begun to converge. But Video Is Not Yet a World This is where the distinction is often misunderstood. For a world model to support meaningful real-time interaction, it must solve several critical problems. Most video models today are essentially answering one question: What should the next frame look like? A true world model needs to answer much more: What happens if I take one step forward? If I walk behind a building and then return, will the building still be there? If I suddenly change the camera angle, will the entire space remain consistent? If I enter a command such as: “Summon a dragon.” Will the world respond immediately? In other words, a world model must do more than generate content. It must understand space. It must understand time. It must understand causality. And it must understand interaction. Moving from watching to participating is where the real difficulty of world models begins. World Models Are Entering the Interactive Era One of the latest attempts in this direction is Alaya World, recently open-sourced by Alaya World, or Alaya Lab. Instead of generating a fixed video clip, it generates a world that users can explore in real time. Users can begin with text, an image, or a video, enter the generated scene, move freely through it, and introduce new prompts at any moment during generation. The world responds immediately. According to the publicly released information, Alaya World provides: Real-time streaming generation at 720p and 24 FPS Stable continuous exploration for more than one minute The ability to switch prompts and trigger skills or events during generation Model weights and inference code released under the Apache 2.0 License Training code and datasets planned for future release What makes these capabilities important is not simply the technical specifications. It is that the generated “world” can now support continuous interaction. The official demo shows that users can genuinely control, transform, and explore the generated environment. AI Is Evolving From a Tool Into an Environment Over the past few years, most discussions around AI have focused on content generation. Generating text. Generating images. Generating videos. But world models raise a fundamentally different question: Can AI generate an environment that people can inhabit, explore, and continuously evolve? If the answer is yes, the impact will extend far beyond video generation. Game development, robotics training, embodied intelligence, digital twins, virtual production, and many other fields could be transformed by the development of world models. World models are still at a very early stage. Yet from Craik’s proposal of an internal mental model more than eighty years ago to the emergence of today’s interactive world-generation systems, a clear evolutionary path is beginning to take shape. Perhaps what AI is ultimately learning has never been limited to images, videos, or language. Perhaps it is learning the world itself. References GitHub: Technical Report:show more

Al‑Qaeda Isn’t Finished. They’re Intent on Turning Past Failures... into Success. One of the most important things people miss about al‑Qaeda is this: They don’t abandon failed plots. They recycle them. Their current U.S. homeland strategy is built on bringing old failures back, and this time, making them succeed. Look at the pattern: They hit the World Trade Center in 1993 and failed to bring it down. Eight years later, they came back and finished the job. The same mindset applies to everything else al‑Qaeda tried and couldn’t pull off: Project Bojinka was the original blueprint for using commercial aviation as a weapon. Parts of it failed in the 1990s. They brought the idea back and used it to build 9/11. They still believe Bojinka must be completed. That is why the aviation plot is back. None of these ideas are over. The group’s current U.S. homeland plot is not new, it is simply being fulfilled by the next generation of al-Qaeda. It is a direct extension of what they attempted before, only now they intend to turn every past failure into a success. Yes, you read that right. Every. Single. One. Abu Bakr Naji, al‑Qaeda strategist and doctrinal advisor, author of The Management of Savagery, believed failure was part of a longer-term strategy, not a terminal endpoint. In his framework, setbacks aren’t the end of jihad, just a step in escalation, refinement, and eventual success. He wrote: “If we fail—we seek refuge with God from that—it does not mean an end of the matter. Rather, this failure will lead to an increase in savagery.” Understanding this is key: the next attack will hit harder. Our intelligence community knows how each plot failed before. So why can’t they thwart them now?show more