<?xml version="1.0" encoding="UTF-8"?><urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:news="http://www.google.com/schemas/sitemap-news/0.9"><url><loc>https://www.gpts24.com/zh/news/mind2act-world-diagnostic-benchmark-finds-robots-cannot-yet-link-reasoning-and-execution</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-22T15:18:07.929Z</news:publication_date><news:title>Mind2Act World 诊断基准：机器人尚无法把推理与执行衔接起来</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/mind2act-world-diagnostic-benchmark-finds-robots-cannot-yet-link-reasoning-and-execution</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-22T15:18:07.929Z</news:publication_date><news:title>Mind2Act World: Diagnostic Benchmark Finds Robots Cannot Yet Link Reasoning and Execution</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/openai-s-roughest-week-gpt-6-2-astra-stumbles-while-anthropic-goes-public</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-22T15:17:15.912Z</news:publication_date><news:title>OpenAI 最难熬的一周：GPT-6.2 Astra 失手，Anthropic 上市</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/openai-s-roughest-week-gpt-6-2-astra-stumbles-while-anthropic-goes-public</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-22T15:17:15.912Z</news:publication_date><news:title>OpenAI&apos;s Roughest Week: GPT-6.2 Astra Stumbles While Anthropic Goes Public</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/open-source-async-rl-framework-bounds-staleness-to-hit-2-68x-throughput-on-llm-post-training</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-22T01:13:56.279Z</news:publication_date><news:title>开源异步强化学习框架限定数据陈旧度，大模型后训练吞吐量最高提升 2.68 倍</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/open-source-async-rl-framework-bounds-staleness-to-hit-2-68x-throughput-on-llm-post-training</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-22T01:13:56.279Z</news:publication_date><news:title>Open-Source Async RL Framework Bounds Staleness to Hit 2.68x Throughput on LLM Post-Training</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/lean-fro-and-mathlib-certify-riemann-hypothesis-proof-four-independent-checks-clear-in-24-hours</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-22T00:46:34.280Z</news:publication_date><news:title>Lean FRO 与 Mathlib 认证黎曼猜想证明：24 小时内通过四项独立核验</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/lean-fro-and-mathlib-certify-riemann-hypothesis-proof-four-independent-checks-clear-in-24-hours</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-22T00:46:34.280Z</news:publication_date><news:title>Lean FRO and Mathlib Certify Riemann Hypothesis Proof: Four Independent Checks Clear in 24 Hours</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/mystery-ai-model-gestalt-tops-contra-labs-design-benchmark-above-all-named-rivals</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-22T00:21:55.185Z</news:publication_date><news:title>神秘 AI 模型 Gestalt 登顶 Contra Labs 设计基准，压过所有具名对手</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/mystery-ai-model-gestalt-tops-contra-labs-design-benchmark-above-all-named-rivals</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-22T00:21:55.185Z</news:publication_date><news:title>Mystery AI Model Gestalt Tops Contra Labs Design Benchmark Above All Named Rivals</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/space-armor-lite-stops-debris-round-behind-body-dummy-what-the-test-shows</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-21T23:58:51.061Z</news:publication_date><news:title>Space Armor Lite 挡住碎片弹丸，身后人体假人无损：测试说明了什么</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/space-armor-lite-stops-debris-round-behind-body-dummy-what-the-test-shows</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-21T23:58:51.061Z</news:publication_date><news:title>Space Armor Lite Stops Debris Round Behind Body Dummy: What the Test Shows</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/two-rival-claims-on-the-riemann-hypothesis-surface-hours-apart-one-built-with-claude-one-without</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-21T05:30:06.889Z</news:publication_date><news:title>两份黎曼猜想证明相隔数小时先后出现：一份借助 Claude，一份没有</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/two-rival-claims-on-the-riemann-hypothesis-surface-hours-apart-one-built-with-claude-one-without</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-21T05:30:06.889Z</news:publication_date><news:title>Two Rival Claims on the Riemann Hypothesis Surface Hours Apart: One Built With Claude, One Without</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/google-s-self-improvement-memory-shows-agent-harnesses</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-21T05:29:21.829Z</news:publication_date><news:title>谷歌自我改进记忆：记下修改理由，智能体 harness 换模型后迁移更好</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/google-s-self-improvement-memory-shows-agent-harnesses</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-21T05:29:21.829Z</news:publication_date><news:title>Google&apos;s Self-Improvement Memory Shows Agent Harnesses Transfer Better When Edit Reasons Are Logged</news:title></news:news></url><url><loc>https://www.gpts24.com/zh/news/anthropic-stock-slides-to-within-5-cents-of-its-83-ipo-price</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>zh-cn</news:language></news:publication><news:publication_date>2026-11-21T04:56:48.847Z</news:publication_date><news:title>Anthropic 股价跌至距 83 美元发行价仅 5 美分，英伟达风波加剧 AI 抛售</news:title></news:news></url><url><loc>https://www.gpts24.com/en/news/anthropic-stock-slides-to-within-5-cents-of-its-83-ipo-price</loc><news:news><news:publication><news:name>GPTS24</news:name><news:language>en</news:language></news:publication><news:publication_date>2026-11-21T04:56:48.847Z</news:publication_date><news:title>Anthropic Stock Slides to Within 5 Cents of Its $83 IPO Price as Nvidia Turmoil Deepens AI Selloff</news:title></news:news></url></urlset>