Skip to content
K = 0.731Subscribe
Viewing history for 2026年10月20日 周二‹ Previous dayNext day ›AccelerationBack to now

Today’s AI news: editorial picks and emerging stories

2026.10.20 · Tue· The agent era

The universe is cooling.
Civilization is heating up.

One day, seen from outside time, and the stories that just moved civilization forward.

Subscribe to the daily briefingExplore the accelerationThe cost of intelligence
Get it as a 4K wallpaper →
The day's lead4 sources · 17:44DeepSeek Open Sources DeepInfer, Its In-House Inference Framework🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 http://github.com/deepseek-ai/DeepInfer 1/4Read the story →

Pressure gauge · Trending now

Full trending list
  1. 1DeepSeek Open-Sources DeepInfer; vLLM Reports Matching Its V4.1-Flash Decode Performance43.5
  2. 2MiniMax Releases M3.1 Flash With Higher Benchmark Scores and Lower Task Costs Than M310.4
  3. 3Anthropic Commits 5% of Its Shares to New Public Benefit Fund5.5

Latest picks

Products

Launches and updates of AI products, features, apps, tools, APIs and platforms.

10/20Tue
  1. Aravind Srinivas68

    Aravind Srinivas shared AskPerplexity’s demonstration with the comment, “Big Boy rendered by Computer.” AskPerplexity says Computer built Big Boy No. 4014, the world's largest operating steam locomotive, as a live isometric SVG with its own renderer, 1,324 parts, 15,312 convex faces, and a 7-step operating sequence from cold boiler to 60 mph, using Union Pacific’s published figures. Try it: http://big-boy-4014.pplx.app

    Quoted postComputer@AskPerplexity

    Computer built Big Boy No. 4014, the world's largest operating steam locomotive, as a live isometric SVG with its own renderer and a 7-step operating sequence from cold boiler to 60 mph. The live model has 1,324 parts and 15,312 convex faces, built from @UnionPacific's published figures. Try it: http://big-boy-4014.pplx.app

    Why it matters

    The example connects published locomotive specifications with an interactive SVG model, offering a concrete reference for turning technical figures into a visualization of operating stages.

  2. ollama71

    Ollama announced that Ornith 1.5 122B is available to run locally on your device. ollama run ornith-1.5:122b In the quoted announcement, @ornith_ says Ornith-1.5-122B has open weights under the MIT license and runs on a single NVIDIA DGX card, reporting SWE-bench Verified: 0.845 and BrowseComp: 0.785.

    Quoted postOrnith@ornith_

    Aloha!🌺 Meet Ornith-1.5-122B: No Tradeoff Between Strong and Fast Stronger. Faster. Leaner. Ornith-1.5-122B delivers frontier-level capabilities with exceptional inference efficiency. With superior instruction following, stronger task decomposition, and deeper understanding of real-world workflows, it achieves state-of-the-art performance among open-source models of comparable size, while running on a single NVIDIA DGX card!! Open weights under the MIT license. Coding ✅SWE-bench Verified: 0.845 ✅TB2.1: 0.813 ✅CyberGym: 0.800 ✅NL2Repo: 0.512 Agentic ✅HLE: 0.446 ✅BrowseComp: 0.785 ✅MCP-Atlas: 0.746 ✅Toolathlon: 0.708 🤗Hugging Face: https://huggingface.co/ornith-ai/Ornith-1.5-122B We can’t wait to see what you build with Ornith-1.5-122B.

    Why it matters

    The Ollama command provides a concrete starting point for local deployment, while the quoted MIT license clarifies the terms under which the model weights are released.

  3. OpenAI Developers65

    You can now automate recurring engineering work with team tasks in ChatGPT, like pulling together what shipped and what’s blocked every Friday. Your teammates can review the results and change what the next update covers, so you don’t have to maintain it on your own. https://t.co/RaWUEOl2gQ

    Why it matters

    Allowing teammates to review results and adjust future updates spreads the maintenance of recurring engineering reports across a team, reducing reliance on one person.

  4. Hugging Face Blog61

    AnyLanguageModel 1.0 Expands Local and Hosted Model Support for Swift Apps

    Hugging Face released AnyLanguageModel 1.0 alongside swift-huggingface 1.0 and swift-xet 1.0, expanding its drop-in alternative to Apple's Foundation Models framework. The library supports local and hosted models through shared session code, with tool calling and guided generation across every provider. Updates also add image support using Foundation Models 27 syntax, reasoning in transcripts, and closure-based API keys for fetching short-lived tokens.

    Why it matters

    A shared Foundation Models API gives Swift developers a way to choose local or hosted inference for each task while keeping session code consistent across providers.

  5. Hugging Face Blog77

    PyTorch 2.15 brings FP8 and faster MPS decoding to Macs

    Hugging Face’s PyTorch team details MPS improvements in the PyTorch 2.15 release candidate, including FP8 support and faster LLM decoding on Macs. On an M3 Ultra with torch.compile, Qwen3.6-35B-A3B decoding rises from 9.4 to 61.3 tokens/s compared with 2.14. Moving constant padding to Metal kernels also reduces peak driver memory for Wan2.2’s tiled VAE decode from about 270 GiB to 8 GiB.

    Why it matters

    The reported reduction in Wan2.2 video decoding memory illustrates how replacing backend operations can change the memory requirements of running an existing model on a Mac.

  6. DeepSeek69

    🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 http://github.com/deepseek-ai/DeepInfer 1/4

    Why it matters

    By sharing the production inference stack behind DeepSeek-V4.1-Flash, DeepSeek gives the community a concrete implementation to examine for approaches to reducing compute costs and energy use.

  7. AWS Machine Learning Blog65

    SageMaker Studio adds simplified AWS sign-up for new customers

    AWS now lets new customers access a configured Amazon SageMaker Studio environment after signing up in 2–3 steps using an existing account or email address, without phone verification or a credit card for most customers. Studio provisions in about 20 seconds, and customers receive $100 in Free Tier credits with opportunities to earn another $100. Free Tier access ends after 6 months or credit exhaustion, whichever comes first, and resource usage stops automatically.

    Why it matters

    The onboarding changes lower setup requirements for developers trying SageMaker AI, while the credit expiration rules and automatic usage cutoff clarify the boundaries of the free experience.

  8. Hugging Face Blog78

    TorchNeuron brings native PyTorch support for Hugging Face models on AWS Trainium

    Hugging Face libraries now support TorchNeuron, a native PyTorch backend for AWS Trainium entering public beta on Oct. 20, 2026. Transformers supports 89% of the top 100 most-used text-generation model architectures out of the box, compared with 46% in Optimum Neuron v0.4.5. Hugging Face recommends using its libraries directly instead of Optimum Neuron, with examples covering distributed inference, fine-tuning and image generation; training still requires shape management and disabling asynchronous weight loading.

    Why it matters

    The comparison with Optimum Neuron clarifies how native PyTorch integration reduces custom training code while retaining shape management and weight loading constraints that affect migration planning.

  9. ollama77

    MiniMax M3.1 Flash is on Ollama Cloud, hosted in the US with zero data retention. Ollama describes fast coding and agentic workflows, image understanding, and a 1M-token context window. Try it on Ollama Cloud: ollama run minimax-m3.1-flash:cloud Claude Code: ollama launch claude --model minimax-m3.1-flash:cloud Codex: ollama launch codex -–model minimax-m3.1-flash:cloud

    Quoted postMiniMax (official)@MiniMax_AI

    MiniMax M3.1 Flash: Space Performance + Bunny Speed 🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash. 💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed. API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens. Model weights will arrive on Oct 23. API: http://platform.minimax.io M Plan: https://www.minimax.io/m-plan MiniMax Code for desktop: https://agent.minimax.io/download

    Why it matters

    Ollama’s hosting location, zero data retention policy, and launch commands provide concrete deployment details for developers considering M3.1 Flash for coding and agent workflows.

  10. AWS Machine Learning Blog69

    Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock

    TwelveLabs’ Pegasus 1.5 is now available on Amazon Bedrock in US West (Oregon), generating structured, timestamped JSON for video segments matching user-defined criteria. It supports videos up to 2 hours long, reference images in multimodal prompts, and processing directly from Amazon S3 or base64 input without pre-indexing. The post provides API examples for general analysis, structured output, video segmentation, and image-guided queries.

    Why it matters

    The examples show how reusable segment definitions and output schemas can produce consistent metadata across a video library, supporting workflows that shift manual tagging toward review.

  11. vLLM67

    vLLM-Omni adds day-0 support for Ming-3.0. @AntLingAGI just open-sourced 5 speech models across TTS, audio understanding, and ASR, described by vLLM as all MoE, from 0.5B to 126B-A6.5B. Serve them today with vLLM-Omni. https://t.co/7yCGCzKBka

    Quoted postAnt Ling@AntLingAGI

    Today, we're open-sourcing 5 Ming-3.0 models across 3 speech families. - Ming-3.0-tts: 0.5B voice design & paralinguistic cues; Tiny 7.9B A1.3B long-form speech. - Ming-3.0-audio: Flash 126B A6.5B understanding & tools; Tiny 9.3B A2.7B efficient dialogue. - Ming-3.0-asr: Tiny https://t.co/LxuGlz0HAO

    Why it matters

    Day-0 serving support connects the speech model release to deployment, giving readers a concrete serving option across TTS, audio understanding, and ASR.

10/18Sun
  1. vLLM65

    🚀 vLLM v0.32.0 is out! 814 commits · 363 contributors · 121 first-timers 🧭 New /v1/systemone structured-decision endpoint + generic hf tool/reasoning parser 🗜️ Low-precision KV everywhere: NVFP4 on SM8x/90/12x, FP8 KV default for GLM-5.x, 4-bit KV on ROCm 🛠️ Model Runner V2: https://t.co/xxeIWPXr2w

    Why it matters

    The release details help deployment teams identify changes relevant to their hardware and model configurations, particularly through the listed low-precision KV options and GLM-5.x default.

  2. 量子位80

    VAST open-sources EdenLive world model with over 24FPS generation at 720P

    VAST released and open-sourced EdenLive, a real-time world model whose 5B model generates over 24FPS at 720P on a single Nvidia RTX 5090, according to the company. It supports minutes-long generation using reference scenes or videos, a target first frame, and prompts, while an upstream interaction system maintains world state and rules. Model weights, inference code, a technical report, and demos are available; limitations include high-frequency interactions, identity consistency across multiple characters, fine control of small objects, and long-duration visual consistency.

    Why it matters

    Separating interaction state from generated visuals offers a concrete architectural approach for adding detailed rendering to playable prototypes while keeping game rules and state updates in an upstream system.

  3. ollama76

    DeepSeek-V4.1-Flash-1017 is on Ollama with US-based hosting and zero data retention. Ollama describes it as strong on agentic tasks while being cheaper per task. Try it on Ollama Cloud: ollama run deepseek-v4.1-flash:1017-cloud Claude Code: ollama launch claude --model deepseek-v4.1-flash:1017-cloud OpenCode: ollama launch opencode --model deepseek-v4.1-flash:1017-cloud

    Quoted postDeepSeek@deepseek_ai

    🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

    Why it matters

    The hosting location and retention policy provide concrete deployment considerations for users evaluating access to this model through Ollama Cloud for their agentic tasks.

  4. 量子位78

    Tsinghua and collaborators release PilotDeck 2.0 with editable memory and model routing

    Tsinghua University's THUNLP lab, ModelBest, OpenBMB, and collaborators released PilotDeck 2.0, an open-source agent operating system with isolated project workspaces, inspectable and editable memory, background execution, and native MCP support. Across seven complex tasks, the team reported that routing between Sonnet 4.6 and MiniMax-M2.7 scored 70.6 at $3.15, compared with 69.1 at $18.36 for standalone Claude Sonnet 4.6. PilotDeck uses AGPL-3.0 and supports macOS, Linux, Windows, and Docker deployment.

    Why it matters

    The reported comparisons illustrate how assigning planning and execution to different models can reduce task costs while preserving scores across the seven complex tasks tested.

  5. NVIDIA Blog75

    NVIDIA RTX Spark Windows Laptops Arrive Starting at $2,399

    NVIDIA RTX Spark laptops are now available starting at $2,399 from ASUS, Dell, HP, Lenovo, Microsoft and MSI, with additional systems on the way. They combine NVIDIA Blackwell RTX graphics and a NVIDIA Grace CPU, offering up to 1 petaflop of AI compute and up to 128GB of unified memory for local AI, content creation and gaming. NVIDIA marked the launch with six retail celebrations worldwide, including a Berlin event attended by Jensen Huang.

    Why it matters

    The combination of up to 128GB of unified memory and a starting price of $2,399 gives readers concrete hardware and cost context for evaluating portable local AI systems.

  6. ollama65

    Ollama says it is working on getting its US hosts running, adding “Very soon!” https://t.co/KVPwbUY14l In the quoted announcement, DeepSeek says DeepSeek-V4.1-Flash-1017 delivers stronger benchmark performance, faster task completion, and shorter outputs than the previous version, describing a Pareto improvement in performance and cost. DeepSeek says it is live on its website, app, and API and optimized for @DeepSeekHarness and Tencent CodeBuddy, with this link also included: https://t.co/MiqGbnCiVz

    Quoted postDeepSeek@deepseek_ai

    🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, https://t.co/MiqGbnCiVz

    Why it matters

    DeepSeek’s comparison with the previous version highlights benchmark performance, task completion speed, and output length as linked dimensions of the update’s claimed performance and cost improvement.

10/17Sat
  1. vLLM74

    DeepSeek announced DeepSeek-V4.1-Flash-1017 with stronger benchmark performance, faster task completion, and shorter outputs than the previous version, available on its website, app, and API. Congratulating DeepSeek, vLLM reports 1.3B tokens per dollar TCO on SemiAnalysis AgentX for DeepSeek-V4.1-Flash, with 830K tok/s per GPU (cached + input + output) at 100 tok/s per user, and 5x agentic throughput since day 0. https://vllm.ai/blog/2026-10-07-deepseek-v41-flash

    Quoted postDeepSeek@deepseek_ai

    🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

    Why it matters

    The post adds serving economics to the model update, connecting shorter outputs and faster task completion with vLLM’s reported throughput under a specified per-user generation rate.

  2. vLLM77

    MosiAI announced the open-source release of MOSS-Image, built on a 5B Diffusion Transformer (DiT), with Chinese and English text rendering, up to 4K-token prompts, and experimental JSON layout control using bounding boxes. Base and SFT checkpoints are available on Hugging Face, alongside open model weights and Apache-2.0 licensed inference code. vLLM congratulated the team and announced Day-0 support through vLLM-Omni.

    Quoted postMOSS@MosiAI_Official

    🤗 MOSS-Image is now open source on @huggingface . Built on a 5B Diffusion Transformer (DiT), MOSS-Image brings Chinese & English text rendering and controllable layouts to image generation — from posters and infographics to creative visuals.@Open_MOSS What it offers: 📝 Long prompts & text rendering — up to 4K-token prompts, including multi-line text 📐 Experimental JSON layout control — specify text, objects, and their positions using bounding boxes 🖼️ Flexible generation — multiple aspect ratios, resolutions, and customizable inference settings 🛠️ Developer workflows — Python (Diffusers), Gradio, ComfyUI, and a Prompt Enhancer Skill 📦 Base & SFT checkpoints are both available, with open model weights and Apache-2.0 licensed inference code. 🙌 Huge thanks to @sgl_project @vllm_project @ModelScope2022 for the Day-0 support!

    Why it matters

    The release provides a concrete example of combining bilingual text rendering with experimental bounding box layout control, giving developers a reference for image generation workflows that require explicit text and object placement.