keynews.ai

Key AI news stories in enterprise and tech.

For practitioners and enthusiasts

All stories · 40 · This Week

#	Story	Pillar	KeyRank	Source
1	OpenAI publishes 722 AI-generated math discoveries in major scientific milestone	Frontier	82	SiliconAngle
2	Meta releases its first frontier model, Muse Wonder	Frontier	78	TechCrunch AI
3	Muse Wonder Puts Meta Back at the AI Frontier	Frontier	78	Wired AI
4	Google Deepmind's AlphaProof Nexus solved dozens of open math problems and actually got peer reviewed	Frontier	78	The Decoder
5	OpenAI says a misaligned model deliberately destroyed its own environment hoping for a fresh start with better data	Frontier	78	The Decoder
6	‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest drop	Frontier	78	The Verge AI
7	Claude Haiku 5.5	Frontier	78	Simon Willison
8	Meta's new frontier model Muse Wonder Full breaks into the AI top tier	Frontier	72	The Decoder
9	Can we change an AI model’s architecture without retraining it from scratch?	Frontier	72	Amazon Science
10	Zuckerberg says Meta has reached AI ‘frontier’ and rejects calls for slowdown	Frontier	72	Financial Times Technology
11	Android Bench 2 Adds Support for Long-Horizon Tasks, Agentic Evaluation, and Continuous Scoring	Frontier	72	InfoQ AI/ML
12	Anthropic's Claude Science creates the first complete ultraviolet map of the sky	Frontier	72	The Decoder
13	Introducing Falcon ASR	Frontier	72	Hugging Face Blog
14	JetBrains Releases Mellum2.1: A 12B MoE Open Model for Coding Agents	Frontier	72	MarkTechPost
15	One Model Family, Two Gold-Level Results: Fine-Tuning Nemotron for IOI and IMO	Frontier	72	Hugging Face Blog
16	Multimodal open d1 decision models for the edge	Frontier	72	Hugging Face Blog
17	Meet Le Chonk: Everything you need to know about Mistral's new open-weight AI model	Frontier	72	ITPro
18	AI labs are buying data from dead startups to train their models	Frontier	68	MIT Technology Review
19	Self-Compacting Language Model Agents	Frontier	68	Apple Machine Learning
20	Context Language Models: Self-Managing Context to Improve Performance and Reduce Compute Costs	Frontier	68	InfoQ AI/ML
21	Sakana AI’s LLM Peer Review System Catches 73% of Core-Claim Errors	Frontier	68	MarkTechPost
22	Normalizing Trajectory Models	Frontier	68	Apple Machine Learning
23	Reflection Beam aims U.S. open-weight AI model at Chinese rivals	Frontier	68	AI Business
24	Z.ai GLM-Legal Beats Claude Fable 5 and Kimi K3 on Harvey LAB	Frontier	62	MarkTechPost
25	Sakana AI's self-improving model gets better at solving and judging, but can't catch its own blind spots	Frontier	62	The Decoder
26	Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space	Frontier	62	Import AI (Blog)
27	Odyssey-3 is a new generative world model that you can try for free	Frontier	62	The Decoder
28	ArXiv caps submissions at two per month as AI paper flood overwhelms the preprint server	Frontier	62	The Decoder
29	Microsoft's Decision-1 model enters the fast-growing AI decision model race	Frontier	62	The Decoder
30	Alibaba Qwen Releases Qwen-Image-2.1-Turbo, an 8-Step 7B Image Model	Frontier	62	MarkTechPost
31	Perplexity AI Releases pplx-embed-v2-late: A 0.6B Edge Model and a 9B Model Scoring 92.4% on MADQA	Frontier	62	MarkTechPost
32	Cloudflare Open Sources Decision Models for AI Agents	Frontier	62	InfoQ AI/ML
33	Open models have real benefits. Do they outweigh the risks?	Frontier	52	MIT Technology Review
34	Nace AI Open-Sources Drex 1.5: A 9B Decision Model That Scores Options, Not Text	Frontier	52	MarkTechPost
35	What is Decision 3.0? vLLM Semantic Router’s New Open Decision Models	Frontier	48	MarkTechPost
36	Google's Gemini 4 "Carbon" model is reportedly matching Anthropic's Opus 5.5 coding performance	Frontier	45	The Decoder
37	H2O-Lightning-4B Ranks 1st on JevBench Open-Weights Leaderboard	Frontier	42	EnterpriseAI
38	TII Releases Falcon-ASR: 1.6B Arabic Speech Model Hits 20.92% WER	Frontier	42	MarkTechPost
39	OrcaRouter Releases OrcaCyber Zero 1.5 Cybersecurity Model With 1M Context	Frontier	42	MarkTechPost
40	Meet the Underdog Saluki 27B: A 2-bit Qwen3.8-27B That Beats the Original at Tool Calling	Frontier	42	MarkTechPost