Hao AI Lab

🚀 We let Jev play Minecraft. And we just can't beat it! 😭 ⚡ Jev: 24 ms decisions 🧠 You: ~200 ms reactions It moves before you even see it move. Way too strong! 🎮 https://mc.alexzms.com (join the server to win)

(1/7) Last week FastVideo released FastH3. Today we bring it local: @NVIDIA DGX Spark 💚 and @Apple Silicon 🍎through MLX. 8x speedup! FastH3 now supported on NVIDIA DGX Spark and Apple Silicon! over base @MiniMax_AI H3 for local generation! This 15 seconds 1344×768 clip was generated on two NVIDIA DGX Sparks. Fully local. @NVIDIARTXSpark @NVIDIAAI

(1/6) Open Weight FastVideo FastH3 V2! Up to 9x speedup on @NVIDIA Blackwell. Lossless quality! 🚀 Lossless quality compared to base 50-step @MiniMax_AI H3. Judge for yourself below! - FastVideo collab w/@nuvalab, NVIDIA FastGen, NVIDIA Enterprise Products - Day 0 @ComfyUI weights and workflows - Day 0 API served on @reactorworld - Omni-ref currently training 👨‍🍳 More comparisons below!

Amazing!! @reactorworld has developed a new infinite stream video using FastH3. The community is moving fast!! And it is fully open source

Your Ray-Ban Meta glasses can already see and hear. Hawky makes it interactive in realtime. Introducing Hawky: an open-source agent that puts realtime interactivity into your phone and your glasses. ☁️Github repo: http://github.com/hao-ai-lab/hawky 🌐Try it now (webapp): http://hawky.live 📱iPhone (beta): http://ios.hawky.live/

Checkout FastVideo's FastH3 Dense served on @vllm_project vLLM-Omni! This is the power of open source! We must win! Thank you guys for the continued support and sponsorship for FastVideo!

Very glad to see this independent evaluation of FastH3 Preview, @MiniMax_AI and H3 Max by @Physion_Labs > FastH3 preview, as a community open-source effort, delivers a decent result overall: it retains strong prompt adherence, including outperforming H3 on that dimension... Stay tuned for our next versions of FastH3 that will be even higher quality!

Check out FastVideo's FastH3 V2 on ComfyUI 🚀🚀

As promised, FastVideo open-sources @MiniMax_AI FastH3 Omni, training code included! 🚀 - Real-time gen @ 12x faster on NVIDIA Blackwell - NVFP4/FP8 for Spark + RTX! - ComfyUI day-0 Collab with @NVIDIAAI , Nuva Lab @nuvalab and Reactor @reactorworld Open source must win! ♥️

FastVideo team excited to announce FastH3-Omni! 🚀🚀 Postrained in collaboration with Nuva Lab @nuvalab and Reactor @reactorworld! The final FastH3-Omni model will be fully open sourced on October 20, along with training code and full support on DGX spark, and RTX GPUs! Please checkout @reactorworld's API access, available today! https://x.com/reactorworld/status/2110064291537953179

GPT and Claude now operate real robot arms. What about open-weight models? 🤔 Introducing ORA: Open Robot-Use Agents. An open harness powered by @deepseek_ai V4.1-Flash, ORA operates robots from visual input—zero-shot, no VLA—and competes with frontier-model agents.

(1/13) Introducing FastVideo UniServe Engine, the fastest serving engine for @MiniMax_AI FastH3 🚀 - 🏭 Production ready: UniServe is serving production API traffic on @reactorworld and @nuvalab on day-0! - ⚡️ Fastest performance: up to 1.8x lower latency and 1.45x higher throughput - 💻 Production Hardware: Supports NVIDIA Blackwell (B200/B300/RTX 6k pro) and Hopper GPUs