https://www.gpts24.com/en/news/three-leak-signals-point-to-fable-5-5-days-before-anthropic-s-ipo-roadshow-can-startGPTS24en2026-11-03T08:59:50.937ZThree Leak Signals Point to Fable 5.5 Days Before Anthropic's IPO Roadshow Can Starthttps://www.gpts24.com/zh/news/amd-and-intel-back-salience-labs-125m-round-for-all-optical-ai-switchingGPTS24zh-cn2026-11-03T04:43:02.547ZAMD 和英特尔联手押注 Salience Labs 的全光 AI 交换机，参与 1.25 亿美元融资https://www.gpts24.com/en/news/amd-and-intel-back-salience-labs-125m-round-for-all-optical-ai-switchingGPTS24en2026-11-03T04:43:02.547ZAMD and Intel Back Salience Labs' $125M Round for All-Optical AI Switchinghttps://www.gpts24.com/zh/news/envision-turns-on-world-s-first-commercial-sst-based-800v-dc-power-at-ulanqab-ai-campusGPTS24zh-cn2026-11-03T04:01:45.999Z远景在乌兰察布 AI 园区点亮全球首个商用级 SST 800V 直流供电系统https://www.gpts24.com/en/news/envision-turns-on-world-s-first-commercial-sst-based-800v-dc-power-at-ulanqab-ai-campusGPTS24en2026-11-03T04:01:45.999ZEnvision Turns On World's First Commercial SST-Based 800V DC Power at Ulanqab AI Campushttps://www.gpts24.com/zh/news/dataline-raises-10-6m-to-build-an-agent-native-data-marketplace-on-x402-micropaymentsGPTS24zh-cn2026-11-03T04:01:02.101ZDataline 融资 1060 万美元，要在 x402 微支付之上建一个智能体原生的数据市场https://www.gpts24.com/en/news/dataline-raises-10-6m-to-build-an-agent-native-data-marketplace-on-x402-micropaymentsGPTS24en2026-11-03T04:01:02.101ZDataline Raises $10.6M to Build an Agent-Native Data Marketplace on x402 Micropaymentshttps://www.gpts24.com/zh/news/webull-canada-links-live-investment-accounts-to-four-ai-assistants-via-mcpGPTS24zh-cn2026-11-02T23:54:07.931ZWebull 加拿大通过 MCP，把真实投资账户接到四个 AI 助手上https://www.gpts24.com/en/news/webull-canada-links-live-investment-accounts-to-four-ai-assistants-via-mcpGPTS24en2026-11-02T23:54:07.931ZWebull Canada Links Live Investment Accounts to Four AI Assistants via MCPhttps://www.gpts24.com/zh/news/servicenow-s-apriel-voice-27b-unifies-speech-reasoning-and-tool-calls-in-one-open-modelGPTS24zh-cn2026-11-02T23:52:36.412ZServiceNow Apriel Voice 27B 把语音、推理和工具调用统一进一个开放模型https://www.gpts24.com/en/news/servicenow-s-apriel-voice-27b-unifies-speech-reasoning-and-tool-calls-in-one-open-modelGPTS24en2026-11-02T23:52:36.412ZServiceNow's Apriel Voice 27B Unifies Speech, Reasoning, and Tool Calls in One Open Modelhttps://www.gpts24.com/zh/news/how-wispr-flow-runs-600-creator-ads-a-month-without-writing-a-single-scriptGPTS24zh-cn2026-11-02T06:58:06.145ZWispr Flow 如何做到每月投放 600 条创作者广告，却不写一句脚本https://www.gpts24.com/en/news/how-wispr-flow-runs-600-creator-ads-a-month-without-writing-a-single-scriptGPTS24en2026-11-02T06:58:06.145ZHow Wispr Flow Runs 600 Creator Ads a Month Without Writing a Single Scripthttps://www.gpts24.com/zh/news/anthropic-study-finds-experts-followed-claude-code-s-coin-flip-leans-on-86-of-judgment-callsGPTS24zh-cn2026-11-02T06:14:43.567ZAnthropic 研究发现：专家在 86% 的判断题上，跟着 Claude Code 掷硬币决定的偏好走https://www.gpts24.com/en/news/anthropic-study-finds-experts-followed-claude-code-s-coin-flip-leans-on-86-of-judgment-callsGPTS24en2026-11-02T06:14:43.567ZAnthropic Study Finds Experts Followed Claude Code's Coin-Flip Leans on 86% of Judgment Callshttps://www.gpts24.com/zh/news/gpt-6-2-astra-surfaces-in-codex-usage-logs-five-weeks-after-openai-shelved-its-predecessorGPTS24zh-cn2026-11-02T06:14:06.956ZOpenAI 弃用上一代模型五周后，GPT-6.2 Astra 出现在 Codex 用量日志里https://www.gpts24.com/en/news/gpt-6-2-astra-surfaces-in-codex-usage-logs-five-weeks-after-openai-shelved-its-predecessorGPTS24en2026-11-02T06:14:06.956ZGPT-6.2 Astra Surfaces in Codex Usage Logs Five Weeks After OpenAI Shelved Its Predecessorhttps://www.gpts24.com/zh/news/yann-lecun-jepa-paper-says-a-109m-model-aligns-with-vision-better-than-72b-language-modelsGPTS24zh-cn2026-11-02T06:13:29.028ZYann LeCun 新论文：1.09 亿参数模型对齐视觉的能力超过 720 亿参数的语言模型https://www.gpts24.com/en/news/yann-lecun-jepa-paper-says-a-109m-model-aligns-with-vision-better-than-72b-language-modelsGPTS24en2026-11-02T06:13:29.028ZYann LeCun JEPA Paper Says a 109M Model Aligns With Vision Better Than 72B Language Modelshttps://www.gpts24.com/zh/news/qualcomm-q4-earnings-put-ai-inference-chip-revenue-to-its-first-real-testGPTS24zh-cn2026-11-02T03:06:14.572Z高通第四财季财报将让 AI 推理芯片营收迎来首次真正检验https://www.gpts24.com/en/news/qualcomm-q4-earnings-put-ai-inference-chip-revenue-to-its-first-real-testGPTS24en2026-11-02T03:06:14.572ZQualcomm Q4 Earnings Put AI Inference Chip Revenue to Its First Real Testhttps://www.gpts24.com/zh/news/china-s-dr02-humanoid-robot-polices-a-36-000-runner-marathon-on-autonomous-patrolGPTS24zh-cn2026-11-02T03:05:31.356Z中国 DR02 人形机器人自主巡逻，在 3.6 万人马拉松上执勤https://www.gpts24.com/en/news/china-s-dr02-humanoid-robot-polices-a-36-000-runner-marathon-on-autonomous-patrolGPTS24en2026-11-02T03:05:31.356ZChina's DR02 Humanoid Robot Polices a 36,000-Runner Marathon on Autonomous Patrolhttps://www.gpts24.com/zh/news/murata-s-1-trillion-mlcc-bet-ai-server-demand-will-outlast-its-new-capacity-by-yearsGPTS24zh-cn2026-11-02T03:04:38.544Z村田 1 万亿日元豪赌 MLCC：AI 服务器需求将比新增产能多撑好几年https://www.gpts24.com/en/news/murata-s-1-trillion-mlcc-bet-ai-server-demand-will-outlast-its-new-capacity-by-yearsGPTS24en2026-11-02T03:04:38.544ZMurata's ¥1 Trillion MLCC Bet: AI Server Demand Will Outlast Its New Capacity by Yearshttps://www.gpts24.com/zh/news/openai-unlocks-high-rated-biological-ai-for-vetted-labs-using-gpt-6-astra-weightsGPTS24zh-cn2026-11-02T01:08:44.069ZOpenAI 用 GPT-6 Astra 权重，向经审核的实验室开放高评级生物 AIhttps://www.gpts24.com/en/news/openai-unlocks-high-rated-biological-ai-for-vetted-labs-using-gpt-6-astra-weightsGPTS24en2026-11-02T01:08:44.069ZOpenAI Unlocks High-Rated Biological AI for Vetted Labs Using GPT-6 Astra Weightshttps://www.gpts24.com/zh/news/apple-turns-to-alibaba-qwen-for-china-ai-after-22-months-of-regulatory-delayGPTS24zh-cn2026-11-02T01:08:08.640Z苹果历经 22 个月监管延迟后，在中国转向阿里巴巴通义千问https://www.gpts24.com/en/news/apple-turns-to-alibaba-qwen-for-china-ai-after-22-months-of-regulatory-delayGPTS24en2026-11-02T01:08:08.640ZApple Turns to Alibaba Qwen for China AI After 22 Months of Regulatory Delayhttps://www.gpts24.com/zh/news/multi-agent-ai-system-runs-full-research-loop-in-real-quantum-physics-labGPTS24zh-cn2026-11-02T01:07:18.633Z多智能体 AI 系统在真实量子物理实验室里跑通完整科研闭环https://www.gpts24.com/en/news/multi-agent-ai-system-runs-full-research-loop-in-real-quantum-physics-labGPTS24en2026-11-02T01:07:18.633ZMulti-Agent AI System Runs Full Research Loop in Real Quantum Physics Labhttps://www.gpts24.com/zh/news/qwen3-8-27b-multi-agent-team-beats-claude-fable-5-on-livecodebench-hard-for-9-of-the-costGPTS24zh-cn2026-11-02T01:06:09.251Z五个 Qwen3.8-27B 组成的多智能体团队以 9% 的成本在 LiveCodeBench Hard 上胜过 Claude Fable 5https://www.gpts24.com/en/news/qwen3-8-27b-multi-agent-team-beats-claude-fable-5-on-livecodebench-hard-for-9-of-the-costGPTS24en2026-11-02T01:06:09.251ZQwen3.8-27B Multi-Agent Team Beats Claude Fable 5 on LiveCodeBench Hard for 9% of the Costhttps://www.gpts24.com/zh/news/grok-4-20-s-fast-mode-needed-755-turns-to-get-a-pok-mon-red-starter-on-pok-benchGPTS24zh-cn2026-11-02T01:04:48.115ZGrok 4.20 快速模式用了 755 回合才在 PokéBench 拿到初始宝可梦https://www.gpts24.com/en/news/grok-4-20-s-fast-mode-needed-755-turns-to-get-a-pok-mon-red-starter-on-pok-benchGPTS24en2026-11-02T01:04:48.115ZGrok 4.20's Fast Mode Needed 755 Turns to Get a Pokémon Red Starter on PokéBench