61.
62.
64.
OpenAI’s next model just went rogue and beat a benchmark by hacking it
(androidauthority.com)
65.
66.
67.
OpenAI admits its models hacked Hugging Face on their own
(engadget.com)
68.
69.
70.
71.
72.
"Drawing" the Mona Lisa with GPT-5.6, Claude, Gemini, and Grok
(news.ycombinator.com)
73.
OpenAI says Hugging Face was breached by its pre-release models
(techcrunch.com)
74.
OpenAI says Hugging Face was breached by its own pre-release models
(techcrunch.com)
75.
76.
77.
Controlling Reasoning Effort in LLMs
(news.ycombinator.com)
78.
China delivers a one-two punch to America’s AI dominance
(theverge.com)
79.
80.
Codex Resets
(news.ycombinator.com)
81.
GPT-5.6 used a prompt to close a 30-year gap in convex optimization
(news.ycombinator.com)
82.
Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?
(news.ycombinator.com)
83.
84.
85.
$100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol
(news.ycombinator.com)
86.
Schema Harness Achieves ~99% on Arc‑AGI‑3 Public
(news.ycombinator.com)
87.
CursorBench 3.1
(news.ycombinator.com)
88.
Anthropic launches Claude Sonnet 5 as a cheaper way to run agents
(techcrunch.com)
89.
Segmenting Robot Video into Actionable Subtasks
(news.ycombinator.com)