{
  "title": "Daily Research Brief 2026-08-10",
  "url": "/en/posts/research-brief-2026-08-10/",
  "permalink": "https://hackcv.com/en/posts/research-brief-2026-08-10/",
  "date": "2026-08-10",
  "lastmod": "2026-08-10",
  "author": "",
  "description": "Daily research brief — AI / LLM / Agent / Computer Vision / Audio-Video / Engineering",
  "categories": ["Research Brief"],
  "tags": ["AI","LLM","Agent","Computer Vision","Audio-Video","Engineering","Daily Brief"],
  "cover": "https://picsum.photos/seed/daily-research-brief-2026-08-10/1200/675",
  "readingTime": 1,
  "wordCount": 171,
  "content": "\u003ch1 id=\"daily-research-brief-2026-08-10\"\u003eDaily Research Brief 2026-08-10\u003c/h1\u003e\n\u003cp\u003e📊 Token usage: estimated from retrieval and writing scale.\u003c/p\u003e\n\u003cp\u003eCovers the latest AI research, open source and industry moves, updated daily.\u003c/p\u003e\n\u003chr\u003e\n\u003ch2 id=\"editors-note\"\u003eEditor\u0026rsquo;s Note\u003c/h2\u003e\n\u003cp\u003eToday\u0026rsquo;s material states a judgment clearly: the breakthrough for long-horizon reliability is shifting from \u0026lsquo;swap in a stronger model\u0026rsquo; to \u0026lsquo;move state out of context\u0026rsquo;. The Horizon Gap surveys 1,547 papers from 2024–2026, with the shared conclusion that outcome-level rewards fail quickly on long tasks; LongHorizon-Harness attacks the same problem from the harness side.\u003c/p\u003e\n\u003ch2 id=\"1-latest-arxiv-papers\"\u003e1. Latest arXiv Papers\u003c/h2\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eThe Horizon Gap: Planning, Memory, Execution, Training, and Evaluation for Long-Horizon LLM Agents\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.06663\"\u003ehttps://arxiv.org/abs/2608.06663\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eLongHorizon-Harness: Advancing Long-Horizon Agents for Real-World Tasks\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.01964\"\u003ehttps://arxiv.org/abs/2608.01964\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eDASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.06243\"\u003ehttps://arxiv.org/abs/2608.06243\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eRemember-R1: Process-Reward-Corrected Multimodal \u0026lsquo;Distributional Visual Forgetting\u0026rsquo;\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.01314\"\u003ehttps://arxiv.org/abs/2608.01314\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eVorch-Omni: Multi-Task Orchestration of Sight and Sound\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.05803\"\u003ehttps://arxiv.org/abs/2608.05803\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eVorch-Streamer: Extending Human Audio-Visual Generation to Real-Time Long-Form Streaming\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.05663\"\u003ehttps://arxiv.org/abs/2608.05663\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eUnderstand Before Detect: Vision-Language Learning for Omni-Domain Infrared Small Target Detection\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.07015\"\u003ehttps://arxiv.org/abs/2608.07015\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003eToken Communication for Multimodal Large Language Model\u003c/strong\u003e — \u003ca href=\"https://arxiv.org/abs/2608.07279\"\u003ehttps://arxiv.org/abs/2608.07279\u003c/a\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n",
  "summary": "Daily Research Brief 2026-08-10 📊 Token usage: estimated from retrieval and writing scale.\nCovers the latest AI research, open source and industry moves, updated daily.\nEditor\u0026rsquo;s Note Today\u0026rsquo;s material states a judgment clearly: the breakthrough for long-horizon reliability is shifting from \u0026lsquo;swap in a stronger model\u0026rsquo; to \u0026lsquo;move state out of context\u0026rsquo;. The Horizon Gap surveys 1,547 papers from 2024–2026, with the shared conclusion that outcome-level rewards fail quickly on long tasks; LongHorizon-Harness attacks the same problem from the harness side.\n"
}