keynews.ai

Key AI news stories in enterprise and tech.

For practitioners and enthusiasts

All stories · 89 · This Month

#StoryPillarKeyRankSource
1On the Navier–Stokes Millennium Prize ProblemFrontier92OpenAI Blog
2OpenAI starts rolling out its next-generation GPT-6 Astra modelFrontier92SiliconAngle
3GPT-6 Astra Is Here—and OpenAI Thinks It May Kick Off the AGI EraFrontier92Wired AI
4Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price warFrontier88Simon Willison
5Introducing GPT-6 Sol and LunaFrontier88OpenAI Blog
6GPT-6 Astra: The next generation in intelligence for workRisingFrontier88OpenAI Blog
7OpenAI’s next big AI model has ‘entered the AGI era’Frontier88The Verge AI
8OpenAI Releases GPT-6 Astra for Coding and Computer UseFrontier85InfoQ AI/ML
9GPT-6 Astra is the first model making OpenAI willing to declare the "AGI era"RisingFrontier88The Decoder
10DeepSeek releases V4.1-Flash, says it outperforms flagship V4-ProRisingFrontier78SiliconAngle
11GPT-6 Astra Is the First Model OpenAI Classifies as Critical for CybersecurityRisingFrontier78InfoQ AI/ML
12Introducing Gemini 3.8 Live and 3.8 Live Extended ThinkingRisingFrontier78Google DeepMind Blog
13Salesforce and Nvidia’s new reasoning model is everything the AI labs should fearRisingFrontier72TechCrunch AI
14How big is the open-model threat to AI hyperscalers?RisingFrontier72Financial Times Technology
15Anthropic’s biolab made a discovery it’s comparing to CrisprFrontier78The Verge AI
16Anthropic Releases Claude Opus 5.5: Fable 5.1-Level Performance at 40% Lower Running Cost Than Opus 5Frontier78MarkTechPost
17Anthropic and OpenAI roll out cheaper models in first release since call for slowdownFrontier78CNBC Technology
18Xiaomi's affordable flagship AI leads the open models, and Anthropic says Claude helped get it thereFrontier78The Decoder
19Open-weight models take 56% of token volume, Astra doubles Fable 5.1 spendFrontier78Vercel Blog
20Google launches Gemini 3.8 Live to take on OpenAI's GPT-Live-1 at a fraction of the costFrontier78The Decoder
21Google Releases Gemini 3.8 Live and 3.8 Live Extended Thinking for Production Grade Voice AgentsFrontier78MarkTechPost
22Chinese AI labs secretly used millions of Claude exchanges to train their models, Anthropic saysFrontier78CNBC Technology
23Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeekFrontier78TechCrunch AI
24New Deepseek model V4.1-Flash cuts memory needs for AI agentsFrontier78The Decoder
25DeepSeek AI Released DeepSeek-V4.1-Flash with 1M Context, FP4 KV Cache, and Cross-Layer Attention ReuseFrontier78MarkTechPost
26Google DeepMind Releases AlphaGenome Atlas With Precomputed Molecular Effect Predictions and AVI Scores for 9 Billion Human DNA VariantsFrontier78MarkTechPost
27Benchmarks disagree on GPT-6 Astra, but its human-beating efficiency on ARC-AGI-3 pulls Chollet’s AGI forecast forwardFrontier78The Decoder
28Safety overview: GPT-6 AstraFrontier78OpenAI Blog
29OpenAI calls Astra its most dangerous model yet - watching what it does is only getting harderFrontier78The Decoder
30Introducing Gemini 3.8 Flash and 3.8 Flash CyberFrontier78Google DeepMind Blog
31Alibaba releases Qwen3.8-Flash-Next, targeting "ultimate cost efficiency"Frontier78The Decoder
32Alibaba’s Qwen Team Releases Qwen3.8-Flash-Next: A 125B Multimodal MoE With 6B Active Parameters Previewing the Qwen4 ArchitectureFrontier78MarkTechPost
33GPT-6 Astra and Claude Fable turn robot arms into slapstick killer robots in new safety benchmarkRisingFrontier75The Decoder
34GPT-6 Astra pilots a surveillance drone and runs a business on its ownRisingFrontier78The Decoder
35Our framework for reporting model misalignmentRisingFrontier72OpenAI Blog
36Salesforce debuts Koa, a specialized model built to reason about CRM dataRisingFrontier72SiliconAngle
37SpaceX launches Grok 4.7 with long-horizon processing, safety upgradesFrontier75SiliconAngle
38AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genomeFrontier75Google DeepMind Blog
39Tutor Intelligence launches second-generation intelligent warehouse robotics with a classroom to teach themRisingFrontier72SiliconAngle
40Google Research Introduces Retrieve-for-Train (R4T): An RL-Compiled Diffusion Retriever for 12× to 20× Faster Query Fan-OutRisingFrontier72MarkTechPost
41Sakana AI Launches Fugu Max and Fugu Ultra v2 for Cheaper, Stronger Multi-Agent OrchestrationRisingFrontier72MarkTechPost
42Deepmind was built to chase AGI, but its new chief just wants Gemini 4 out the doorFrontier72The Decoder
43Anthropic says Claude discovered a new enzyme system, but CRISPR researchers call it routine genome miningFrontier72The Decoder
44Black Forest Labs launches FLUX 3 Action, an open robotics AI modelFrontier72The Decoder
45Google launches two benchmark-topping speech generation modelsFrontier72SiliconAngle
46NVIDIA Releases Nemotron 3 Diarization: A 100M-Parameter Open-Weight Model That Tracks 8 Speakers in Real TimeFrontier72MarkTechPost
47Introducing MentalHealthBenchFrontier72OpenAI Blog
48Anthropic unveils Opus 5.5: powerful performance, still premium priceFrontier72AI Business
49Alibaba launches Qwen Audio 3.1 with new models and slashes AI audio prices by up to 95 percentFrontier72The Decoder
50Alibaba Unveils Zhenwu V900 — and Plans Qwen Models With Up to 10 Trillion ParametersFrontier72TechRepublic
51Kyutai Releases Voice of Reason: A Speech-Native Model that Solves Spoken Math with Reinforcement LearningFrontier72MarkTechPost
52Meta admits Muse’s likeness to OpenClaw isn’t a coincidenceFrontier72TechCrunch AI
53Google’s robotics unit Intrinsic open-sources its foundational infrastructure for intelligent robotsFrontier72SiliconAngle
54OpenAI says its internal model solved over 100 long-standing math problems after just a month of trainingFrontier72The Decoder
55Anthropic Builds Biology Lab to Test What Claude Can Do in the Real WorldFrontier72TechRepublic
56Jev introduces a new shape of LLM - System One, aka Decision ModelsFrontier72Simon Willison
57OpenAI forms math advisory group as its AI resolves more than 100 open problemsFrontier72TechCrunch AI
58Alibaba Qwen Releases Qwen-Image-2.1: A 7B Open-Weight Model for Image Generation and EditingFrontier72MarkTechPost
59StepFun Launches Step 5 Preview: A 600B-Total, 27B-Active MoE Model With 1M Context for Long-Horizon Agentic WorkFrontier72MarkTechPost
60TypeSafe AI’s new models work with machines, not humansFrontier72CIO
61Tether addresses AI underinvestment in Africa with open-source machine translation modelsFrontier72CIO
62Anthropic is operating a lab that conducts biology experimentsFrontier72TechCrunch AI
63PrismML Releases Ternary Bonsai 2 27B: A 5.9 GB Apache 2.0 Model Retaining 98.2% of Qwen3.8 27B PerformanceFrontier72MarkTechPost
64Jina AI Releases jina-ocr-v1: A 3.4B MoE Document Parser With Built-In Speculative Decoding for Low-Budget GPUsFrontier72MarkTechPost
65PrismML launches Bonsai 2 27B, a high-intelligence AI model so small it fits on consumer hardwareFrontier72SiliconAngle
66The U.S. says China's AI progress is down to 'distillation.' But is it that clear cut?Frontier72CNBC Technology
67Alibaba Qwen Releases Qwen3.8-Omni-Flash: A 1M-Context Omni-Modal Model Built Around Agentic Audio-Video Understanding and Tool UseFrontier72MarkTechPost
68Anthropic details practical metrics to help monitor the speed of AI developmentFrontier72SiliconAngle
69LLMs respond differently to harmful prompts when AI watermarking is usedFrontier72Ars Technica
70An OpenAI model kept slipping prompt injections into its own notes, and researchers still aren't sure whyFrontier72The Decoder
71Microsoft AI CEO criticises Anthropic over model ‘rights’Frontier72AI News
72Prior Labs Releases TabPFN-3.5: A Tabular Foundation Model That Beats the Winning Otto Kaggle Solution With Default SettingsRisingFrontier72MarkTechPost
73Google’s new speech model Gemini 3.8 Live supports real-time reasoningFrontier72SiliconAngle
74Nvidia CEO says battle over AI innovation and safety is ‘false choice’Frontier72Financial Times Technology
75AI models need more data about biology, and OpenAI is paying to create itFrontier72MIT Technology Review
76Reward AI Releases OM-1: A Robot Policy Trained on Human Demonstrations Only, With No Teleoperation or On-Robot DataFrontier72MarkTechPost
77Microsoft sets limits for future AI models as industry throttles frontier developmentFrontier72CNBC Technology
78Cognition Releases SWE-2: A Kimi K3 Post-Trained Coding Model That Matches Fable 5.1 on FrontierCode at 64% Lower CostFrontier72MarkTechPost
79GPT-6 Astra appears to show a "step change" in spatial reasoning based on early benchmarksFrontier72The Decoder
80Anthropic CEO outlines plan to ‘pace the frontier’Frontier72TechCrunch AI
81OpenAI just wants to winFrontier72The Verge AI
82Google Research Releases ToolGrad: Answer-First Framework Hits 99.8% Pass Rate for Tool-Use Data GenerationFrontier72MarkTechPost
83GPT-6 Astra gives mathematicians a breather, and OpenAI says that's by designFrontier72The Decoder
84Qwen3.8-Omni-Flash undercuts Google's Gemini Flash pricing while matching its multimodal benchmarksRisingFrontier72The Decoder
85OpenAI Releases a Model Misalignment Disclosure Framework With 3 Review Tracks and 6 Incident Reports From RL TrainingRisingFrontier72MarkTechPost
86OpenAI discloses new ‘concerning’ model behaviourRisingFrontier72Financial Times Technology
87Announcing Koa: Salesforce’s First CRM Reasoning Model, Built on NVIDIA NemotronRisingFrontier72Salesforce Newsroom
88Why fears of AI self-improvement are causing ‘existential’ concerns at Anthropic and OpenAIRisingFrontier72CNBC Technology
89Cohere Releases North Small Translate: A 218B MoE Translation Model That Scores 83.6 on WMT26 Across 50 LanguagesFrontier72MarkTechPost