🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4

👋 Hello, X. We’re Physis. AI’s next frontier is the physical world. We’re building a general-purpose world foundation model—one that can accumulate and reuse experience across scenarios to understand, predict, and act. 🎯 Our goal is general physical intelligence, with applications across embodied AI, industrial simulation, and scientific prediction. This brings us to a foundational question: 🤔How can we build a foundation for physical intelligence that accumulates and reuses experience across scenarios—and understands, predicts, and influences the real world at the speed and accuracy each task requires? Over the coming months, we’ll share our thinking, models, and systems as we work toward this goal. Today, we begin with infrastructure. ✨ Together with vLLM, we’re introducing PhysisServe. Born at UC Berkeley’s Sky Computing Lab, vLLM now powers inference on an estimated 500,000 GPUs worldwide at any given time. Together, we’re advancing efficient, scalable serving for the next frontier: physical AI. Three core designs make this possible: • Stage-wise pipelining across denoising steps and Transformer blocks. • Time-sharing and batching to improve GPU utilization. • Virtual ranks and joint scheduling to share resources across streams. 📊 To explore how world action models (WAMs) and action-conditioned world models (ACWMs) can sustain higher output rates as they scale, we benchmark PhysisServe on five backbones, including MiniMax H3 33B. On 200 GPUs, this 33B backbone delivers a chunk every 4.083 ms—about 245 chunks/s.* The figure below shows how additional compute enables higher sustained output rates across model sizes. *DiT denoising only; excludes VAE decoding. Output interval is not end-to-end response latency. 👉 Read our first blog post on infra: physis-ai.com/blog/physisser… 👉 Explore the code: github.com/Physis-AI/PhysisS…

Introducing d3-turbo: our newest state-of-the-art multimodal foundation decision model⚡ #1 on both Jev Decision Index boards. It points(box), follows and sees images in your questions. StateOnce brings 15x faster than d3 flash 🤗 huggingface.co/vllm-sr/d3-tu… #vLLM #Jev #Decision

Today, we're open-sourcing 5 Ming-3.0 models across 3 speech families. - Ming-3.0-tts: 0.5B voice design & paralinguistic cues; Tiny 7.9B A1.3B long-form speech. - Ming-3.0-audio: Flash 126B A6.5B understanding & tools; Tiny 9.3B A2.7B efficient dialogue. - Ming-3.0-asr: Tiny 8.6B A2B dialects & context.

Speculators v0.9.0 is out. Training got faster and lighter: a fused loss kernel gives up to 2x faster forward training passes on small (~1B) models, and an FP8 hidden-state backend cuts storage 50% while keeping speculator quality. Full notes: github.com/vllm-project/spec…

Introducing Decision 3.0: towards open multimodal foundation decision models ⚡ Open and state-of-the-art from 0.6B to 27B: #1 on both Jev Decision Index boards in our evaluation, on a new Pareto frontier. HF collection 🤗: huggingface.co/collections/v… #OpenSource #Jev #LLM

Fresh vLLM containers can recompile thousands of Python files at startup. One build flag moves that work to image build time: 14–18% less startup time (~18 s) in my H100 benchmarks. Already in the CUDA nightly; official release images from v0.32.0. vllm.ai/blog/2026-09-22-pyth…