{"slug": "physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops", "title": "PhysBrain 1.5: Chinese Physical AI Breakthrough - Physical Foundation Model Tops Open Source Rankings", "summary": "Chinese company DeepCybo released PhysBrain 1.5, a physical foundation model built on a unified 'Physical Loop' architecture that scores 72.5 across 28 public benchmarks, ranking first among open-source models and just behind closed-source leaders GPT-6 Astra (73.3) and Gemini 3.6 Flash (73.0). The Apache 2.0-licensed model was trained on Ego360, a human panoramic video dataset capturing full-body pose, hand movement, and task-level voice, and drew over 3,000 Hugging Face downloads within three days of launch.", "body_md": "Physical AI is becoming the most critical track in global AI for 2026. While large models have spent three years pushing \"can talk, can think, can work\" to the extreme, the industry is now asking: **Can models enter the real world? Can they take action?**\n\nRecent experiments have put this question front and center:\n\nBut top LLMs still have a gap in physical manipulation. On millimeter-level precision tasks like \"blue puzzle piece alignment insertion,\" GPT-6 Astra's success rate drops from 95% to 10%.\n\nOn September 9, 2026, Chinese company **DeepCybo** (深度机智) released **PhysBrain 1.5**, a physical foundation model that achieves **72.5 average score across 28 public benchmarks** — ranking **#1 among open-source models**, just 1 point behind top closed-source models GPT-6 Astra (73.3) and Gemini 3.6 Flash (73.0).\n\n| Metric | Result | \n|---|---|\n| Open Source Rank | #1 | \n| Average Score (28 benchmarks) | 72.5 | \n| Open Source #1 | 14 benchmarks | \n| Open Source #2 | 10 benchmarks | \n| vs. Hy-Embodied-VLM-1.0 (66.0) | +6.5 points | \n| vs. RynnBrain 1.1 (63.1) | +9.4 points | \n\nBoth models are fully open-source with Apache 2.0 license.\n\nPhysBrain 1.5 is built around a unified architecture called **Physical Loop**:\n\nThis is a continuous, stable closed-loop system that self-corrects based on feedback.\n\nDeepCybo's key insight: **Physical intelligence cannot be achieved by translating internet text into actions.** Humans grow operational skills through repeated \"see — try — get feedback — correct.\" \n\nThe team built **Ego360**, a human panoramic real-data system, using full-body pose, hand movement, and task-level voice from panoramic video. Physical pre-training supervision comes entirely from human interaction videos.\n\nWithin 3 days of launch, PhysBrain 1.5 received **3,000+ downloads** on Hugging Face.\n\nPhysBrain 1.5 demonstrates that physical foundation models are not a short-term trend but a long-term strategic investment. By focusing on human learning patterns and building a complete data-model-body loop, DeepCybo has positioned PhysBrain at the forefront of global physical AI.\n\n*Tags: AI, Robotics, PhysicalAI, OpenSource, DeepLearning*", "url": "https://wpnews.pro/news/physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops", "canonical_source": "https://dev.to/ryan_zhao/physbrain-15-chinese-physical-ai-breakthrough-physical-foundation-model-tops-open-source-51ok", "published_at": "2026-09-15 02:04:12+00:00", "updated_at": "2026-09-15 02:31:13.076089+00:00", "lang": "en", "topics": ["artificial-intelligence", "robotics", "ai-research", "ai-startups", "ai-products"], "entities": ["DeepCybo", "PhysBrain 1.5", "GPT-6 Astra", "Gemini 3.6 Flash", "Ego360", "Hugging Face", "Hy-Embodied-VLM-1.0", "RynnBrain 1.1"], "alternates": {"html": "https://wpnews.pro/news/physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops", "markdown": "https://wpnews.pro/news/physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops.md", "text": "https://wpnews.pro/news/physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops.txt", "jsonld": "https://wpnews.pro/news/physbrain-1-5-chinese-physical-ai-breakthrough-physical-foundation-model-tops.jsonld"}}