#LLM
Curated coverage on «LLM»: releases, research, benchmarks, open-source.
Microsoft Releases MAI-Cyber-1-Flash: 5B Active Parameters and 95.95% on CyberGym
Microsoft AI has released MAI-Cyber-1-Flash, a 5B-active-parameter cyberdefense model that, paired with the MDASH harness, lifted its CyberGym score to 95.95%.
Apple-π: A Benchmark That Tests Whether Video Models Understand Physics or Just Copy the Picture
The new Apple-π benchmark grades the physical reasoning process of video models rather than the final frame: do they understand Newton's laws, or just copy visual…
Ontologies Are Back: How AI Agents and Neo4j Are Reviving the Semantic Web
AI engineers are rediscovering ontologies, structured knowledge graphs, to keep probabilistic LLM agents inside deterministic logical boundaries.
OpenAI study: how ChatGPT expands the scope of work and blurs the lines between professions
OpenAI Gives 100,000 Researchers Free Access to Its Most Advanced ChatGPT Models
OpenAI: Two API Settings Tripled GPT-5.6's ARC-AGI-3 Score — to 38.3%
OpenAI's GPT-5.6 Bets on Efficiency: More Intelligence per Dollar
LTM to Roll Out Cognition's Devin to 260 Clients, Aiming to Clear 80% of CVE Backlogs
LTM and Cognition launched BlueVerse RightLogic: Devin will fix vulnerabilities for banks and enterprises, with a target of closing 80% of a CVE backlog, up from…
DeepSeek Pauses 480 Billion Yuan Funding Round After Founder's Remarks Leak
DeepSeek founder Liang Wenfeng verbally asked investors to hold off on the round after his private comments leaked and went viral on Chinese social media.
OpenAI Flagged GPT-5 as Dangerous: The Model Produced Recipes for Poisons and Bioweapons
GPT-5.6 Generated a Proof That Solved a 6-Year-Old Cryptography Problem
DeepSeek Founder Liang Wenfeng: What the Leak of His Closed-Door Meeting Revealed
Cursor: Cheap Models Write Code to a Frontier Model's Plan, Cutting Cost 15x
Anthropic Releases Claude Opus 5 — Close to Fable 5 in Capability and Stronger at Coding
Anthropic has released its flagship Claude Opus 5, which the company says approaches Fable 5's capabilities and is notably stronger at complex programming tasks.
Anthropic Unveils Opus 5: Cheaper and Less Restricted Than Fable
Anthropic has released Opus 5, a new flagship Claude model. Per TechCrunch, it is cheaper and less restricted in its responses than Fable, and so preferable in most…