https://megadose.ai/i/9e5afa38-4378-461a-a3e9-5654e29a35b7 Megadose en 2026-10-16T17:50:01+00:00 Three CoreWeave co-founders have sold $3.3B in stock since its March 2025 IPO, including $2.4B in 2026 alone; CoreWeave stock fell 60%+ from its June 2025 peak (Dylan Sloan/Bloomberg) https://megadose.ai/i/0956e6d6-e9f1-463a-9cec-3dda81975daa Megadose en 2026-10-16T16:40:02+00:00 Reflection AI releases Apache 2.0-licensed weights for Beam, a 501B-parameter model, trained with RL that penalized unnecessary tokens (Matt Marshall/VentureBeat) https://megadose.ai/i/449bf075-a1ff-4af8-88fd-f1dbd7227c9e Megadose en 2026-10-16T16:01:37+00:00 Elorian’s first AI model wants to reason with images, not just about them https://megadose.ai/i/bdd1ff1f-8e68-4609-bc6f-54d2e7d36ece Megadose en 2026-10-16T12:45:02+00:00 Waymo plans to launch its robotaxis in Toronto in 2028 in partnership with Element Fleet Management, the company's first deployment in Canada (Prathik Jayaprakash/Reuters) https://megadose.ai/i/32cfe01a-f888-452b-8c87-e1c649d6cf81 Megadose en 2026-10-16T04:55:00+00:00 Taiwan plans US industrial parks costing $2B to $4B each, starting with a Texas complex costing up to $4B, to build a supply chain for Nvidia, TSMC, and Foxconn (Bloomberg) https://megadose.ai/i/4a603d5f-011a-42f4-aca9-d828088bbde6 Megadose en 2026-10-16T03:40:00+00:00 Austin-based RapidCanvas, which uses agents to help businesses build custom AI tools, raised a ∼$29M Series B led by Vertex Ventures and Hitachi Ventures (Chris Metinko/Axios) https://megadose.ai/i/9535ca7b-430b-428b-b572-60d22d00b611 Megadose en 2026-10-16T03:15:02+00:00 Sources detail Nscale's setbacks ahead of an IPO at up to a $35B valuation, including a six-month delay in Norway and penalties owed to Microsoft and ByteDance (Bloomberg) https://megadose.ai/i/cc2a01fd-c069-49c1-b217-3fc833577864 Megadose en 2026-10-16T02:20:01+00:00 Microsoft unveils MAI-Base-2, a MoE model with 2.8T total and 80B active parameters, claiming 4x-12x more pretraining FLOPs-efficiency than DeepSeek-V4-Pro-Base (Microsoft AI) https://megadose.ai/i/8ee211c2-c052-40de-8e6f-e1665dd8b213 Megadose en 2026-10-16T01:30:03+00:00 OpenAI continues to work to grow its ad business in ChatGPT, cutting ad rates, chasing small and midsize businesses, and hiring some new talent from Meta (Alix Coutures/The Information) https://megadose.ai/i/3516242a-8261-4e15-b018-d154e3dd670f Megadose en 2026-10-16T00:15:02+00:00 Opsin, which helps companies show where AI agents are active, what data they access, and flag risky behavior, raised $21M led by SYN Ventures and Race Capital (Stephen Council/Business Insider) https://megadose.ai/i/dede22f1-5c7c-4b21-b277-04483f26e4c1 Megadose en 2026-10-15T21:35:04+00:00 Bluesky expands Attie, its AI tool for building custom feeds, with a "social vibe coding" feature to create sites, games, quizzes, and more on the AT Protocol (Sarah Perez/TechCrunch) https://megadose.ai/i/47a27d46-a6db-4198-a613-194e33815395 Megadose en 2026-10-15T21:20:01+00:00 Sources: some Anthropic investors discussed a ~$1.5T valuation for its IPO, down from ~$2T they had expected, after disclosures showed slowing Q3 revenue growth (Reuters) https://megadose.ai/i/627fb80c-89b5-431c-a310-7f82b51ae0b3 Megadose en 2026-10-15T20:36:10+00:00 A researcher says modern neural networks are far more interpretable than believed, with most neurons understandable in models up to trillions of parameters (Nick Cammarata/Goodfire) https://megadose.ai/i/967e67f0-d48f-4d87-8d9e-89251c6ab6c7 Megadose en 2026-10-15T20:04:53+00:00 Meta launches an open-weight AI model that runs on Macs https://megadose.ai/i/b2bf42c8-8b76-408f-80cf-9200da36f1a0 Megadose en 2026-10-15T20:00:13+00:00 Perplexity open sources SPACE, its AI agent sandbox, and says partners including CoreWeave, Akamai, and Crusoe agreed to offer SPACE to customers (Reed Albergotti/Semafor) https://megadose.ai/i/3ab54202-95c2-49a9-afff-5986aaa98f0f Megadose en 2026-10-15T19:00:00+00:00 Bringing GPT-Rosalind to academic research labs https://megadose.ai/i/e68f6728-3491-48cd-a8ba-4795847dfe7b Megadose en 2026-10-15T18:56:46+00:00 Bluesky expands its AI-powered Attie app to let anyone build websites, games, and more https://megadose.ai/i/4d06921b-3d91-49c1-a54d-bad831c1550e Megadose en 2026-10-15T17:59:59+00:00 LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation https://megadose.ai/i/b091e786-6369-4225-bf22-721e4fec101e Megadose en 2026-10-15T17:59:20+00:00 Discovering state-of-the-art quantum circuits with large language models https://megadose.ai/i/d4b29651-8cf8-4dfc-a37d-7de82fdc2b99 Megadose en 2026-10-15T17:59:11+00:00 T2-z: Test-Time Optimization of Environment Latents for Robotic World Models https://megadose.ai/i/58a8e63f-c108-4408-9925-aa7bd9192208 Megadose en 2026-10-15T17:59:02+00:00 CoDA: Co-Evolution of Data Engines and Action Policies https://megadose.ai/i/f8d1743a-5b88-4f3d-a87a-0c61f3e426a4 Megadose en 2026-10-15T17:58:45+00:00 PathFM: Fast Generative Forecasting of Multivariate Trajectories https://megadose.ai/i/47cac691-b023-4f2f-b8df-24a605a76963 Megadose en 2026-10-15T17:58:04+00:00 OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling https://megadose.ai/i/5d070f71-8f50-4ee9-8caf-bb75e182ce86 Megadose en 2026-10-15T17:58:02+00:00 Rethinking Routing Exploration in Modern Mixture-of-Experts https://megadose.ai/i/6f4b2027-1b43-456f-925a-4460ca00cd74 Megadose en 2026-10-15T17:57:23+00:00 FDE-Gym: The Last Mile of AI Delivery https://megadose.ai/i/01ae74b2-7c08-4712-8be6-efa09958411d Megadose en 2026-10-15T17:56:32+00:00 MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering https://megadose.ai/i/451b7dba-8308-4d40-989a-2c5fbf75da3f Megadose en 2026-10-15T17:56:18+00:00 GeoState: Learning View-Decoupled Geometry States for 3D World Modeling https://megadose.ai/i/59f852a3-607a-4205-b5f8-86bd7784ffdf Megadose en 2026-10-15T17:56:07+00:00 3DCodeVerse: Exploring 3D Worlds as Code https://megadose.ai/i/75cd1e0b-a56e-47fb-befd-221205107bdc Megadose en 2026-10-15T17:55:50+00:00 Dexterity from Supervised Play https://megadose.ai/i/7bcf0208-afb0-400e-b10f-556f7cadc1c8 Megadose en 2026-10-15T17:55:21+00:00 Lean Graph Search for Formal Theorem Proving https://megadose.ai/i/967df784-1620-413a-baf4-f14c4a4cd744 Megadose en 2026-10-15T17:55:01+00:00 Volvo says it will use Nvidia's Hyperion platform to launch a Level 2++ hands-free, point-to-point automated driving system by 2030 (Andrew J. Hawkins/The Verge) https://megadose.ai/i/d604334f-5f35-4e96-b0e0-d4d6fe671978 Megadose en 2026-10-15T17:54:13+00:00 Robust Reinforcement Learning for Dynamic Treatment Regimes under Model and Measurement Uncertainty https://megadose.ai/i/ff835906-cdee-4aa5-b46f-d4ea660fa80e Megadose en 2026-10-15T17:53:33+00:00 Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? https://megadose.ai/i/99f25f6b-82cf-4bd7-82ac-ee883e50382d Megadose en 2026-10-15T17:52:33+00:00 KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents https://megadose.ai/i/73ccc058-7bb9-4760-a0b7-5228a4e39bcf Megadose en 2026-10-15T17:52:31+00:00 Designing Transformers for Continuous Diffusion Language Models https://megadose.ai/i/2750e04c-c375-4d15-93fb-ba08ca35bf26 Megadose en 2026-10-15T17:52:24+00:00 ACEdit: Stable Multi-Turn Image Editing via Asymmetric Conditioning https://megadose.ai/i/0f500f6c-d2e7-4489-8586-229f35bf8c54 Megadose en 2026-10-15T17:52:18+00:00 Twice Upon a Time: Complementary Representations for Time-Series Reasoning https://megadose.ai/i/cb69aaba-c8b8-4d99-844e-6802045209cd Megadose en 2026-10-15T17:52:12+00:00 Training Reasoning Models to Generate Diverse Scientific Ideas https://megadose.ai/i/b0de6674-3fb6-4324-b110-0379dac19a8d Megadose en 2026-10-15T17:51:57+00:00 Learning Self-Orchestration in Decentralized Agent Swarms https://megadose.ai/i/9915f0b8-9228-4fba-811d-5838e1a238b8 Megadose en 2026-10-15T17:51:56+00:00 AVC-Reward: Event-Centric Audio-Video Consistency Evaluation https://megadose.ai/i/f80af58d-24cd-4ed3-9047-87d3e5f33478 Megadose en 2026-10-15T17:51:32+00:00 A team of Gemini 4 Argon agents beats Minecraft https://megadose.ai/i/728b9bb6-48dd-4c57-ae99-d394af1027da Megadose en 2026-10-15T17:51:17+00:00 Learning to Steer LLM Agents and Multi-Agent Systems https://megadose.ai/i/77fa8d79-7917-4fb1-b560-2d9030aa00c2 Megadose en 2026-10-15T17:51:01+00:00 MindEval2: Clinically Grounded Patients and Rubrics for Evaluating Language Models in Mental Health Support https://megadose.ai/i/62646184-bfdd-453b-a3c6-f2128f1fcc8e Megadose en 2026-10-15T17:50:57+00:00 Grid-Bench: Can LLMs Predict How Discrete Worlds Change? https://megadose.ai/i/910926c3-be0f-42f7-b9d2-e92e4405f45b Megadose en 2026-10-15T17:47:52+00:00 Knowledge That Is Linear Before Fine-Tuning Generalizes After It https://megadose.ai/i/aeb45316-06e4-43e1-b012-7220a01558e5 Megadose en 2026-10-15T17:46:37+00:00 AI Consilium: a Benchmark of LLM Clinical Multidisciplinary Second Opinions https://megadose.ai/i/aab6a6de-eeca-49fe-8fb9-0d2663ceb435 Megadose en 2026-10-15T17:45:51+00:00 Private Retrieval Augmented Generation via Random Projection https://megadose.ai/i/f4d0f9ff-a38d-4870-8ce7-e4e0ef672f8e Megadose en 2026-10-15T17:45:27+00:00 Backward Compatibility in Zero-Bit Neural Audio Watermarking https://megadose.ai/i/700a3a92-3a28-4635-b7a9-a04c497a69ff Megadose en 2026-10-15T17:44:46+00:00 Multi-Target Few-Shot Prompting with Diffusion Language Models https://megadose.ai/i/663e40f2-03c2-4134-b42a-86e81defd7e2 Megadose en 2026-10-15T17:44:43+00:00 Towards Continual Learning: Mitigating Forgetting With Subspace Tuning https://megadose.ai/i/042f622d-19d0-4f79-b982-9f9e3387deda Megadose en 2026-10-15T17:43:47+00:00 CAT-alyzing Honesty: Continuous Adversarial Training for Transferable LLM Honesty https://megadose.ai/i/f5ea72d2-93b4-4000-9af9-4f84e5cc2810 Megadose en 2026-10-15T17:43:31+00:00 EliteReplay: Preserving Diverse Solutions in RL Post-Training via Quality-Diversity Search https://megadose.ai/i/9035b21b-80d0-4a5a-b21b-4e14557c511e Megadose en 2026-10-15T17:43:16+00:00 Improving Environment-Agnostic Coding Agents: The Role of Scaffolding https://megadose.ai/i/fdd4e7ee-841c-4f56-a2a7-889d7fdb4c0a Megadose en 2026-10-15T17:42:52+00:00 Learning Rational Strategies for Temporal Objectives in Multi-Agent Systems https://megadose.ai/i/51705197-1748-4a23-85b8-47ace0710238 Megadose en 2026-10-15T17:41:46+00:00 Asking Smarter Questions: Targeted Active Preference Learning for Underspecified Rewards https://megadose.ai/i/dd4a1dc0-ff44-48d4-a242-00a8dde89f42 Megadose en 2026-10-15T17:40:12+00:00 LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents https://megadose.ai/i/7324ed02-6c28-4511-8fa0-434fe49960d1 Megadose en 2026-10-15T17:35:17+00:00 Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries https://megadose.ai/i/a5e80e03-19b8-48e9-a3be-cdf9d8894f2c Megadose en 2026-10-15T17:34:17+00:00 X-RIC: An Agentic RAN-Intelligent Controller Benchmark for Enhancing Reinforcement Learning https://megadose.ai/i/37cc14ea-26ed-4f29-82b1-9977d67b9630 Megadose en 2026-10-15T17:28:50+00:00 Baselines and Score Centering in Policy Gradient Estimation https://megadose.ai/i/7541bfdf-ed73-4148-a7ac-566037da8e5b Megadose en 2026-10-15T17:28:39+00:00 Topology-Inspired Fine-Tuning for Wind-Field Super-Resolution https://megadose.ai/i/a843039e-af0b-46f3-9bff-c0a62bae91f9 Megadose en 2026-10-15T17:24:51+00:00 BitSparse: Compact Activation Storage for ReLU Feed-Forward Networks https://megadose.ai/i/09d6eae0-3cee-48f0-92f5-57070cc6b04b Megadose en 2026-10-15T17:22:59+00:00 Explore Once, Evaluate Many: Multiple Policy Evaluation for Stochastic Linear Bandits https://megadose.ai/i/d591f95d-d924-4748-ac38-0c3811ce951c Megadose en 2026-10-15T17:22:11+00:00 Not X, but When: Diagnosing a Mandarin LLM Writing Tic through Contextual Appropriateness https://megadose.ai/i/173bcd55-4ebe-4566-bfee-465ff3166a49 Megadose en 2026-10-15T17:20:27+00:00 Meta says it worked with Apple to make Muse Spark Open run locally on M5 Ultra Mac Studios, letting users pair two to four Studios to code or do agentic work (CNBC) https://megadose.ai/i/89e05299-38a1-4d36-8598-dacb4a2bb2b0 Megadose en 2026-10-15T17:19:57+00:00 Disagreement-Guided Weak-to-Strong Routing https://megadose.ai/i/4c42572d-bd49-4bd2-9a22-35c37a897e90 Megadose en 2026-10-15T17:19:40+00:00 Language Is a Security Surface for Agent Guards https://megadose.ai/i/87f9e0c6-0f30-4e90-9417-6260253be3c5 Megadose en 2026-10-15T17:18:05+00:00 Shortcut Sensitivity in Agent-Safety Evaluations https://megadose.ai/i/4b533353-4c2f-44aa-84c2-de181e5ac9a3 Megadose en 2026-10-15T17:16:56+00:00 SurvICL: a Foundation Model for in-context Survival Analysis https://megadose.ai/i/906791b8-a209-4a57-a892-3e23d706856f Megadose en 2026-10-15T17:15:52+00:00 Query-Efficient Rejection Sampling for Parallel Masked Diffusion https://megadose.ai/i/c6837782-2f7f-43cb-b4e8-5ae8b16c97b5 Megadose en 2026-10-15T17:10:01+00:00 Meta releases Muse Spark Open, saying it is the most capable US open-weight model, with 816B total and 42B active parameters and a 1M-token context window (Meta AI Research) https://megadose.ai/i/c131c846-4d3b-45e9-8343-5b09464e56cc Megadose en 2026-10-15T16:38:42+00:00 Krea launched Krea 3 for fast and precise image editing https://megadose.ai/i/adec52e1-f28d-49ea-8b79-c3d39999f6da Megadose en 2026-10-15T16:20:26+00:00 Nvidia launches Nemotron 3.5 Super VL, a 120B-parameter open model for enterprise agents, and says it outperforms frontier models on specialized benchmarks (Carl Franzen/VentureBeat) https://megadose.ai/i/37751a00-99f5-4461-b830-fa34716a354c Megadose en 2026-10-15T16:04:05+00:00 Krea 3 – An image model built for editing at scale https://megadose.ai/i/dab22481-58bc-4a0f-b37d-e13596b41dd0 Megadose en 2026-10-15T14:29:33+00:00 Google’s Universal Cart shopping tool launches in the US https://megadose.ai/i/c41db0d0-b516-480e-a9c8-f7644fa81080 Megadose en 2026-10-15T13:55:00+00:00 Portland-based Mueon, which says its AI server racks could fit up to 10x more computing capacity, raised $65M led by Maverick Silicon, with Intel participating (Phoebe Liu/The Information) https://megadose.ai/i/5ce4e488-3882-4337-98d7-eacac25dc837 Megadose en 2026-10-15T13:30:02+00:00 North.cloud, which helps companies watch their cloud, AI and data spend, raised a $60M Series B led by PeakSpan Capital, taking its total funding to $67M (Chris Metinko/Axios) https://megadose.ai/i/7c895db9-766c-488e-9299-d43dea06a263 Megadose en 2026-10-15T13:04:03+00:00 Anthropic tells investors quarterly revenue up nearly 50% to $16.8bn ahead of IPO https://megadose.ai/i/4bab08f8-8dfe-4452-b23c-762bffe4aece Megadose en 2026-10-15T12:10:01+00:00 Amazon launches two AI wearables, the $75 Bee Frontier wristband and $100 Bee Eclipse magnetic phone-attachment recorder, with a $12/month Bee Pro subscription (Bloomberg) https://megadose.ai/i/fee651af-a47e-4a6e-8a4f-a20859464323 Megadose en 2026-10-15T12:00:00+00:00 Amazon’s Bee launches two new personal AI devices, starting at $75 https://megadose.ai/i/8abc5e5e-d736-49cd-b213-1b51dbc01d45 Megadose en 2026-10-15T12:00:00+00:00 Introducing Timestral https://megadose.ai/i/7d9b38a9-30d9-4372-88ee-b1739322bbcc Megadose en 2026-10-15T10:08:08+00:00 [AINews] Modal’s $1B ARR, $16B Series D makes it NYC’s newest AI Infra Decacorn, 3x usage and 3x valuation in 5 months, 15x in 1 yr https://megadose.ai/i/d33a8abf-2baf-4b0a-a6e9-b908d41c0951 Megadose en 2026-10-15T09:40:01+00:00 TSMC says it plans to build an additional 3nm fab in Tainan, Taiwan, set for production in H1 2028, alongside its 3nm capacity in Arizona and Japan's Kumamoto (Qianer Liu/The Information) https://megadose.ai/i/15c09e1a-02ab-40b7-9fa6-d9f286a1e5f2 Megadose en 2026-10-15T07:47:04+00:00 Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL https://megadose.ai/i/d990d2ba-032b-4bbf-a93b-f41199c8b150 Megadose en 2026-10-15T06:20:01+00:00 Figma announces new features to define how AI agents work on its platform, including Figma Specs, a Markdown editor that lets users define their design language (Ivan Mehta/TechCrunch) https://megadose.ai/i/168e21e7-bad8-4497-86f0-3e1e2c6c75e9 Megadose en 2026-10-15T05:56:47+00:00 TSMC reports Q3 revenue up 50.9% YoY to $46.36B, net income up 68.7% to $23.7B, both above est., marking a sixth straight quarterly profit record on AI demand (Jenny Lee/CNBC) https://megadose.ai/i/714c772a-d808-4d35-8cc2-7534ef5b77e0 Megadose en 2026-10-15T04:35:04+00:00 Bloomberg is making its ASKB conversational AI interface available to all Terminal users, letting them integrate their own content with existing financial data (Steven Rosenbush/Wall Street Journal) https://megadose.ai/i/7cf01dfe-3dc2-4bde-8d3b-f27eb69a33ed Megadose en 2026-10-15T04:30:00+00:00 Figma announces new features to define how agents work on its platform https://megadose.ai/i/2c9c076c-a34c-4b34-a4c2-e2a5e6e6a9bf Megadose en 2026-10-15T03:45:01+00:00 Sources: ByteDance is in advanced talks to buy data center ASICs from MediaTek, potentially becoming its second major customer for such chips after Google (Bloomberg) https://megadose.ai/i/39c38be2-6ac8-405d-b9f9-6d953b7c34d7 Megadose en 2026-10-15T03:40:00+00:00 Sources: Anthropic told investors that distillation accounts for 5% to 7% of revenue; its rapid but slowing revenue growth could weigh on its IPO valuation (Cory Weinberg/The Information) https://megadose.ai/i/0e19b2f7-3cc0-4d94-b03f-6a933118acb4 Megadose en 2026-10-15T00:00:00+00:00 Learning Self-Orchestration in Decentralized Agent Swarms https://megadose.ai/i/48f7a86a-d2a8-4d30-a08c-ffad41782045 Megadose en 2026-10-15T00:00:00+00:00 Mosaic: Multi-Session Agentic Reinforcement Learning https://megadose.ai/i/45ed565e-312c-455a-8a6e-44616705f9ec Megadose en 2026-10-15T00:00:00+00:00 CarryOn: A Dual-Loop Harness for Personalized and Reliable Robot Assistance https://megadose.ai/i/748a4f51-2fcb-477a-b9f7-50e260e27deb Megadose en 2026-10-15T00:00:00+00:00 Trajectory-Level Speculation with Semantic Verification for Tool-Using Agents https://megadose.ai/i/404beaa4-1856-4293-acfb-39a2c83f6a78 Megadose en 2026-10-15T00:00:00+00:00 EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment https://megadose.ai/i/44a77f7b-b0a7-4610-bf4c-f6fbdb307a76 Megadose en 2026-10-15T00:00:00+00:00 Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models https://megadose.ai/i/5fcc4876-3d3f-4ea9-aae8-4253bd53ea5e Megadose en 2026-10-15T00:00:00+00:00 Routing Policy Optimization for Online Diffusion Reinforcement Learning https://megadose.ai/i/49be2a05-8ec4-4fae-8c9e-c06133f58191 Megadose en 2026-10-15T00:00:00+00:00 Velox: Spatial Memory Makes World Models Efficient https://megadose.ai/i/aa1b67af-eba4-45fc-af02-53ced48e91f6 Megadose en 2026-10-15T00:00:00+00:00 Memory as Weights: Internalizing Long-Term History for Streaming Videos https://megadose.ai/i/d0a16231-4262-4395-a46b-712cd622261f Megadose en 2026-10-15T00:00:00+00:00 A team of Gemini 4 Argon agents beats Minecraft https://megadose.ai/i/c6b35945-6bd6-4d20-abd0-806beb5e84c8 Megadose en 2026-10-15T00:00:00+00:00 Athanor: An Agent-Native System for Autonomous Post-Training Research https://megadose.ai/i/06528245-2e52-4f2c-aca8-69a4f3d14958 Megadose en 2026-10-15T00:00:00+00:00 TimeRSI: Towards Self-Improved Agentic Deployment for Time-Series Foundation Models https://megadose.ai/i/3a2da4b4-b016-4c7b-ace4-67348fe1583a Megadose en 2026-10-15T00:00:00+00:00 Gen4P: Video Generators are 4D Perceivers https://megadose.ai/i/115c1832-a19f-489b-90b9-49cd7124c127 Megadose en 2026-10-15T00:00:00+00:00 When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill https://megadose.ai/i/c94ffe7e-829e-4089-8ba9-c2f973be3be0 Megadose en 2026-10-15T00:00:00+00:00 SUPRA-TEX: 3D Generation with Super-Aligned Texture https://megadose.ai/i/659e52de-62a7-4011-8ce9-8567bbd8d2cd Megadose en 2026-10-15T00:00:00+00:00 Can World and Action Modeling Benefit Each Other under Unified Causal Training? https://megadose.ai/i/83200aef-468f-4525-bd09-6f89f137e1a3 Megadose en 2026-10-15T00:00:00+00:00 FDE-Gym: The Last Mile of AI Delivery https://megadose.ai/i/a1eaba2e-8f94-40ae-9da2-ed06fa6862fe Megadose en 2026-10-15T00:00:00+00:00 Data World: Scaling Data Environments from Scratch https://megadose.ai/i/40f7d09a-2231-4757-82e9-24ee3fd35f0c Megadose en 2026-10-15T00:00:00+00:00 OmniVerifier-C1: Decomposed Graded Verifier for Rule-Governed Generative World Modeling https://megadose.ai/i/17b95eb8-a644-4e25-a80d-c8e3323a378c Megadose en 2026-10-15T00:00:00+00:00 CoWorld Reasoner: Agentic Multimodal Reasoning with the Evolving Video World Model-Driven Harness https://megadose.ai/i/38ecb712-5bfb-4712-aa9d-a2c2f721d4dc Megadose en 2026-10-15T00:00:00+00:00 OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use https://megadose.ai/i/d85fa453-eb4a-43ba-b14d-20220dbd45e5 Megadose en 2026-10-15T00:00:00+00:00 CoDA: Co-Evolution of Data Engines and Action Policies https://megadose.ai/i/14f24278-e91a-4851-8289-736e1eab8997 Megadose en 2026-10-15T00:00:00+00:00 MJev-Spatial: System One Spatial Intelligence via Native Attention Matching https://megadose.ai/i/11ce7a24-e79d-4e6e-9121-fa674664b0c1 Megadose en 2026-10-15T00:00:00+00:00 VLA-Δ: Learning Physical State Transition for Vision-Language-Action Models https://megadose.ai/i/b93683e1-37ef-475b-b285-84189350633f Megadose en 2026-10-15T00:00:00+00:00 Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation https://megadose.ai/i/c1b2d58b-5c47-4130-ab5a-2937f041a1e7 Megadose en 2026-10-15T00:00:00+00:00 MMSearch-OPD: Easy Made Hard, Hard Made Learnable for Long-Horizon Multimodal Search Agents https://megadose.ai/i/0c4a85bb-ba84-4b6b-91af-b29bb38e92c1 Megadose en 2026-10-15T00:00:00+00:00 MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering https://megadose.ai/i/d8200018-f3e7-4a40-ad97-bed80f56f7e0 Megadose en 2026-10-15T00:00:00+00:00 PathFM: Fast Generative Forecasting of Multivariate Trajectories https://megadose.ai/i/892eadcf-4681-43ac-af5e-a9af0c9b555f Megadose en 2026-10-15T00:00:00+00:00 Explaining the Roles of Experts in Mixture-of-Experts LLMs for Personalization https://megadose.ai/i/89d4c321-c345-4bcd-8757-dd375ea2800b Megadose en 2026-10-14T23:59:37+00:00 LFM2.5-24B-A2B: Built for fast on-device agents https://megadose.ai/i/0aeb6a54-8d31-449e-8780-58f674b36af9 Megadose en 2026-10-14T22:30:03+00:00 Sources: Anthropic told investors that its revenue jumped to $16.8B in Q3, up from $11.5B in Q2, and that the company was profitable on an adjusted basis in Q3 (Financial Times) https://megadose.ai/i/c593fd33-bc05-4089-a197-2277aee545e1 Megadose en 2026-10-14T21:40:01+00:00 Zenzap, a private messenger with AI agents that aims to replace apps like WhatsApp in the workplace, raised $44M from Bessemer, Lightspeed, and others (Ivan Mehta/TechCrunch) https://megadose.ai/i/ea2fc042-f353-46e9-a255-b08c54cd11cf Megadose en 2026-10-14T19:25:03+00:00 SpaceXSI debuts Grok Voice 3.0, saying it tops Artificial Analysis' speech-to-speech index, responds in under a second, and costs less than rivals, at $0.08/min (SpaceXSI) https://megadose.ai/i/d6566001-db9d-45f3-aa8c-43354d83f8db Megadose en 2026-10-14T18:32:52+00:00 Banks are now trying to offload the AI data center debt they just underwrote