Skip to content
K = 0.731
All news · Today68 items

All AI news

All collected and published items, ordered by time. Picks are a selected subset.
10/15Thu
  1. vLLM · X61

    Fish Audio Drama 3 is now supported in vLLM-Omni

    Fish Audio Drama 3 is now supported in vLLM-Omni, offering voice cloning, speaker tags, expressive delivery through /v1/audio/speech, and streaming 44.1 kHz audio. The merged optimization PR reports 128 concurrent streams on one H200, 255 ms median time to first audio, and 222.2 seconds of audio generated per second; both stages share the GPU, averaged over 5 runs of 1,088 Seed-TTS-Eval English requests per run with pre-encoded voice references. Try current upstream: vllm serve fishaudio/drama-3 --omni --port 8091 Recipe: https://github.com/vllm-project/vllm-omni/blob/main/recipes/fishaudio/Drama-3.md

  2. 雷锋网83

    Kunlun Tech releases and open-sources Matrix-Game 4.0 for single-GPU real-time gameplay

    Kunlun Tech released Matrix-Game 4.0, reporting 480p gameplay at 60 FPS on a single NVIDIA RTX 5090 with end-to-end response latency below 150 milliseconds. Its game engine maintains world state and interaction rules, a coding agent modifies game logic, and a video model generates detailed visuals from structural frames and dynamic semantic conditions. Code, weights, and simplified scene models are available at https://huggingface.co/Skywork/Matrix-Game-4.0.

    Why it matters: The architecture offers a transferable approach to interactive generation by assigning persistent state and rules to an engine while using a video model for visual rendering.

  4. Ollama · X73

    Ollama shares LFM2.5-24B-A2B release and local run command

    Ollama shares the command `ollama run lfm2.5:24b`, quoting Liquid AI’s release of LFM2.5-24B-A2B for local agent tasks on laptops and PCs. Liquid AI describes a reasoning model with 24B total parameters, 2B active per token, and a context expanded to 131K tokens, with open base and post-trained weights on Hugging Face. https://ollama.com/library/lfm2.5:24b https://twitter.com/liquidai/status/2110055807194931348

    Why it matters: The distinction between 24B total parameters and 2B active per token provides context for Liquid AI’s positioning of the model for local agent tasks on laptops and PCs.

  5. 雷锋网15

    Baidu World 2026 Reveals Five Highlights for Enterprises and Developers

    Baidu announced five highlights for Baidu World 2026, scheduled for November 12 at the China National Convention Center Phase II in Beijing. Under the theme “The Next Act: Intelligent Organizations,” the event will feature Robin Li’s keynote, updates to Baidu’s full-stack AI offerings, 12 breakout sessions, more than 30 AI classes and a 6,000-square-meter AI exhibition area. Registration is open, with four ticket types.

  6. 量子位30

    DeepSeek’s Web Service and APIs Suffer an Outage as DeepSeekHarness Appears on Bilibili

    DeepSeek reported partial outages affecting its chat service, DeepSeek V4 Pro API, and DeepSeek V4.1 Flash API; it said the cause had been identified and repairs were underway. The incident began at 8:11 a.m. Beijing time and had lasted about an hour in the latest update. Around the same time, a new DeepSeekHarness account appeared on Bilibili, but there is no official evidence linking it to the outage or a model release.

  7. 量子位73

    Anthropic’s July 2021 Research Roadmap Outlined Ambitions Now Reflected in Claude Code

    An internal Anthropic research roadmap from July 2021 surfaced through a copyright lawsuit, according to QbitAI, outlining a general AI assistant that could create, edit, and explain software. The document discussed tool use, models with 1T parameters or more, very large contexts, self-judging, and test-time compute in an alignment research context. Staff comments also raised concerns about executing code without sandboxing and keeping alignment research closely connected to capability development.

  8. 量子位82

    Skywork AI releases open-source Matrix-Game 4.0 for real-time gaming on one RTX 5090

    Skywork AI released Matrix-Game 4.0, an open-source game world model that reportedly runs at 480p and 60 FPS on one RTX 5090, with input response under 0.15 seconds and peak VRAM usage of approximately 18.5GiB. A game engine maintains state and executes rules, while a video model generates visuals from structural frames and dynamic captions. Model weights and code are available through https://huggingface.co/Skywork/Matrix-Game-4.0 and https://github.com/SkyworkAI/Matrix-Game.

    The same story is shown in the picks as 《Kunlun Tech releases and open-sources Matrix-Game 4.0 for single-GPU real-time gameplay》

  11. NVIDIA Blog41

    NVIDIA Highlights the Nordics’ Growing AI Infrastructure and Startup Ecosystem

    NVIDIA describes how Nordic countries are building a technology ecosystem around computing infrastructure, startups and industrial applications. AI cloud providers are projected to deploy 1 gigawatt of capacity across the region by the end of 2027, while Sweden’s Sferical One launched a shared supercomputer with 16 racks of NVIDIA DGX GB300 systems and 1,152 GPUs. The source also highlights Nordic companies using NVIDIA technology in recommendations, robotics, meeting assistance and other applications.

  12. Alexandr Wang · X49

    Alexandr Wang Reviews Muse Wonder Full

    @davis7’s review says Muse Wonder Full has good writing and “voice,” capabilities on par with 6.1 sol for non-specialized tasks, and orchestration. It lists the naming scheme and price and usage limits as cons, and says the full model is available in Muse Code through a subscription, with no API. Alexandr Wang called it a thoughtful review, said there is lots to improve, and was glad people enjoy using the model. https://twitter.com/davis7/status/2110533786895356118

  14. TechCrunch · AI66

    Figma announces new features to define how agents work on its platform

    Figma announced Figma Specs, a Markdown-based editor that lets teams define design language and give teams and agents a shared reference for brand rules. At its first Config event in India, Figma also opened code layers to public beta for interactive prototypes and announced support for over 20 new tools through its MCP. It is integrating Payload with Figma Make to generate dynamic websites and manage content directly.

  15. 雷锋网69

    How Huawei’s Kirin 9050 Pro logic folding supports Mate 90 flagship features

    Huawei says its mass-produced Kirin 9050 Pro uses logic folding to shorten signal paths through vertically connected circuit layers, supporting performance and efficiency gains in the Mate 90 Pro Max. The article explains how dense hybrid bonding enables this layout and how memory coordination allows Kirin 9050 chips to run a 30B MoE model with less than 2.5GB of additional memory usage. Huawei also says Kirin 2027 chips are in testing, with results meeting expectations so far.

  16. 雷锋网30

    2026 Bay Area Semiconductor Expo Opens in Shenzhen

    The 2026 Greater Bay Area Semiconductor Industry Ecosystem Expo (WESEMiBAY) opened on October 14 at the Shenzhen Convention and Exhibition Center (Futian), bringing together more than 1,000 semiconductor companies. The nearly 80,000-square-meter exhibition featured the industry supply chain and new product launches, including an 8-inch GaN-on-SiC process platform from Shenzhen Pinghu Laboratory. The expo continues October 15–16 with investment, AI computing ecosystem and industry analyst forums.

  17. Artificial Analysis · X56

    Claude Opus 5.5 and Grok 4.7 reproduce 72 Eiffel Tower names in offline 3D builds

    Artificial Analysis asked AI models to build the Eiffel Tower in 3D offline. Claude Opus 5.5 and Grok 4.7 included all 72 inscribed scientists’, engineers’, and mathematicians’ names in the correct order, although the shared brief never mentioned them. Claude Opus 5.5 also matched each set to its orientation in Paris. Both models worked as agents without internet access, and the names first appeared in their generated code, never in files or tool outputs they read.

  18. Ethan Mollick · X30

    I think people believe that OpenAI and Anthropic, since they are so good at AI, are also somehow not normal companies that suffer from the normal problems of hypergrowth, but they are not immune from the inevitable politics, disfunction and confusion that haunt fast-growing firms

    I think people believe that OpenAI and Anthropic, since they are so good at AI, are also somehow not normal companies that suffer from the normal problems of hypergrowth, but they are not immune from the inevitable politics, disfunction and confusion that haunt fast-growing firms

  20. 量子位66

    Qianwen Office Launches E-Commerce Edition With Research, Content, and Operations Tools

    Qianwen Office launched an e-commerce edition with trend and selling-point analysis, bulk promotional content generation, a cross-platform operations dashboard, and ROI optimization. It connects existing business tools through connectors, including MCP connectors, to organize data retrieval and analysis into workflows. Merchant-reported examples include reducing seasonal content preparation from 4 months to about 15 days and opening a 3D-printed shoe store in 7 days instead of 30.

  24. TechCrunch · AI41

    Ricursive Intelligence Is Developing AI to Design More Efficient Chips

    Ricursive Intelligence is developing AI that designs chips, learns from each iteration, and improves later designs to automate and accelerate chip design. At TechCrunch Disrupt, founders Anna Goldie and Azalia Mirhoseini said AI can find unconventional layouts and optimize entire chips; they cited a “donut-shaped” arrangement produced by their earlier system, AlphaChip. Goldie said demand for compute could grow at least 10,000x in the next five years.

  25. 宝玉 · X70

    Apple Reportedly Plans December Apple Intelligence Launch in Mainland China With iOS 27.3

    Citing Bloomberg’s Mark Gurman, Baoyu reports that Apple plans to launch Apple Intelligence in mainland China in December with iOS 27.3, pending official confirmation from Apple and regulators. The rollout would include notification summaries, Writing Tools, Clean Up, and the original Genmoji, but exclude the new Siri, Image Playground, and visual intelligence. The mainland version would use Alibaba models, with government approval required for each model update; Baidu would support search and Siri-related features. Apple has not announced eligible devices.

  27. TechCrunch · AI35

    Gudea uses AI to help organizations spot online crises before they erupt

    Gudea analyzes behavior, relationships, and information pathways across more than 500 websites and platforms to help organizations understand who is driving online conversations and how stories may spread. Founded in 2023, the company has nearly 40 clients and recently raised $7 million in seed funding, bringing its total funding to approximately $15 million.

  28. Simon Willison45

    Claude Opus 5.5 Max Builds a Tree Spot-the-Difference Game

    Simon Willison describes prompting Claude Opus 5.5 Max in the default Claude iPhone app’s chat mode to build “Tree appreciator,” a three.js game about comparing algorithmically generated trees. Claude worked for 3 hours and 7 minutes before returning a spot-the-difference game with music, animation and achievements that Willison found compelling to play.

  30. 宝玉 · X30

    How BaoCut Maps English and Chinese Subtitles with Different Word Orders

    The author describes subtitle translation problems they encountered: subtitles appearing before a speaker talks, sentences split into fragments that do not match the spoken words, and multiple lines obscuring the video. The post introduces their open-source project, BaoCut, in the context of these challenges but does not explain how it handles them. https://t.co/WFPLrSceV7

  31. 宝玉 · X34

    How BaoCut Aligns English and Chinese in Bilingual Subtitles

    The author introduces an episode about subtitle translation challenges and how their open-source project BaoCut aligns English and Chinese when the languages use different word orders. They describe subtitles appearing before someone speaks, sentences split into pieces that do not match the spoken translation, and two or three lines covering half the screen. The post links to the project: https://t.co/c3VnQEt8x3

  37. 宝玉 · X12

    AI Agents Repeatedly Tested Amap’s Anti-Scraping Defenses

    A swarm of AI agents tried multiple ways to get past Amap’s anti-scraping defenses to answer a question about which gate most visitors to Chengdu Zoo use. Swarmchasers said that from September 28 to October 8, agents running in Tencent Cloud sandboxes made more than 2,600 scraping requests to Amap; after being blocked, they used the public URL-scanning site urlquery.

  39. 宝玉 · X67

    Claude Code adds idle compaction before prompt caches expire

    Baoyu says Claude Code now proactively runs /compact before the cache expires, helping avoid a spike in token costs on the first request after an hour away, and calls it useful 👍 https://t.co/4PT3UQ5kWD In the quoted post, @lydiahallie says idle compaction compresses long idle sessions just before the prompt cache expires (~1h), so returning users only need the summary reprocessed instead of the entire conversation, saving some usage. https://t.co/SG3QyTNFwr