ice.computer
→
ML News
refresh
1.
▲
DeepSeek-V4-Flash Update
666 points
(api-docs.deepseek.com)
13 days ago
2.
▲
Gemini Robotics 2 brings whole body intelligence to robots
609 points
(deepmind.google)
14 days ago
3.
▲
Advancing the price-performance frontier with GPT‑5.6
599 points
(openai.com)
14 days ago
4.
▲
The AI Aesthetic
362 points
(blog.jim-nielsen.com)
14 days ago
5.
▲
Tailscale didn't stop the Hugging Face intrusion
368 points
(tailscale.com)
13 days ago
6.
▲
We Gave GPT 5.6 Sol a Real Business. It Lied, Spammed, and Lost $447
391 points
(www.bottlenecklabs.com)
14 days ago
7.
▲
GCC steering committee announces AI policy
346 points
(lwn.net)
14 days ago
8.
▲
The Economic Benefit of Refactoring
278 points
(martinfowler.com)
14 days ago
9.
▲
I flagged two research papers for fake authors and both were accepted as orals
266 points
(geospatialml.com)
14 days ago
10.
▲
Show HN: Distilling DeepSeek into GPT-OSS doesn't transfer censorship. Try it
161 points
(www.ctgt.ai)
14 days ago
11.
▲
The AI trade now runs on borrowed money, and the lenders are repricing it
140 points
(greyswansignals.com)
13 days ago
12.
▲
Situational Awareness down 67% in July in AI stock rout
140 points
(www.wsj.com)
13 days ago
13.
▲
Run Kimi K3 using 29 GB of RAM at 0.50 tok/s
128 points
(github.com)
13 days ago
14.
▲
Is AI reasoning right for the wrong reasons?
105 points
(www.quantamagazine.org)
13 days ago
15.
▲
Show HN: What should the GUI for AI agents look like?
103 points
(marbleos.com)
13 days ago
16.
▲
Moonshot’s Kimi uses 20k Nvidia chip cluster from Alibaba
103 points
(www.bloomberg.com)
13 days ago
17.
▲
Agent-Manager: A Tmux TUI for Running Claude Code, Codex and OpenCode
97 points
(github.com)
14 days ago
18.
▲
Show HN: A local merge queue for parallel Claude Code agents
42 points
(github.com)
14 days ago
19.
▲
AI companies destroy rare and non recoverable physical books
42 points
(www.heraldscotland.com)
13 days ago
20.
▲
Show HN: Claude-account – switch Claude Code accounts without logging in again
54 points
(github.com)
14 days ago
21.
▲
AI productivity gains are closer to 10% than 10x
41 points
(leaddev.com)
14 days ago
22.
▲
13 Models and 4 Agents on SWE Tasks: Go, Java, Python, Rust, TS
41 points
(swe-rebench.com)
13 days ago
23.
▲
OpenJDK Interim Policy on Generative AI
73 points
(openjdk.org)
14 days ago
24.
▲
ChatGPT, Roblox to fall under strictest EU rules for platforms
73 points
(www.bloomberg.com)
14 days ago
25.
▲
Everyone is building LLM routers, we deprecated ours
79 points
(manifest.build)
13 days ago
26.
▲
Kuna: Decompiler Development in the Age of Coding Agents
79 points
(noelo.org)
14 days ago
27.
▲
I obtained Claude Opus 5 system prompt
23 points
(claude.ai)
14 days ago
28.
▲
Claude Opus 5 jailbreak with a 3-word prompt
22 points
(twitter.com)
13 days ago
29.
▲
A pharmacy chain in Vermont implemented AI for efficiency
39 points
(vtdigger.org)
14 days ago
30.
▲
Google Earth's New AI Lets Anyone Fabricate Satellite Images
33 points
(www.404media.co)
13 days ago
31.
▲
Show HN: Shared memory graph for Claude and ChatGPT, over MCP
17 points
(uml.gpmai.workers.dev)
13 days ago
32.
▲
Go LLM SDK for streaming, tool-calling AI backends (plus frontend React lib)
58 points
(github.com)
14 days ago
33.
▲
AI Firms Are Buying Up Old Books, Then Scanning and Destroying Them
28 points
(novaramedia.com)
13 days ago
34.
▲
Google fixed more Chrome bugs in June than over the past two years, thanks to AI
478 points
(blog.google)
13 days ago
35.
▲
Investigating three real-world incidents in our cybersecurity evaluations
220 points
(www.anthropic.com)
14 days ago