-
$2.7B to bring him back, gone in 20 months: Noam Shazeer leaves Google DeepMind for OpenAI
This isn’t a product launch — it’s a hire that moves the frontier. On June 18, Noam Shazeer announced he’s leaving his job as Gemini co-lead at Google DeepMind to join OpenAI. Sam Altman called it a hire “10 years in the making.” The number that stings: Google paid a reported $2.7B back in 2024… Continue reading
-
Anthropic accuses Alibaba’s Qwen lab of the largest-ever distillation attack (28.8M Claude queries)
This isn’t a product launch — it’s an accusation, and it landed in a letter to US senators and the White House. Anthropic says Alibaba’s Qwen lab spun up nearly 25,000 fake accounts and fired off 28.8 million Claude interactions between April 22 and June 5, 2026, then used the outputs to train Qwen. What… Continue reading
-
360 Tulongfeng & Yitianzhen — China’s cyber-AI agents fill the export-ban gap
The US banned non-Americans from Anthropic’s Mythos and Fable 5 on June 12. Two weeks later, Asia is building around the wall. What 360 actually shipped At ISC.AI 2026 in Beijing, 360 Security unveiled two AI agents. Tulongfeng hunts software vulnerabilities — point it at code and it finds exploitable flaws automatically. 360 claims it… Continue reading
-
Anthropic Mythos 5 — U.S. government clears limited release
Anthropic’s strongest model is back online — but only for a list the government wrote. On June 26 the Trump administration cleared Anthropic to release Mythos 5 to roughly 100 companies and federal agencies. That’s two weeks after Commerce ordered both Fable 5 and Mythos 5 pulled offline for every user on Earth, citing a… Continue reading
-
Mistral OCR 4 wins 72% of head-to-head tests — and tells you where it’s unsure
Mistral dropped OCR 4 on June 23, an open-weights document-understanding model that does more than turn PDFs into text. It returns bounding boxes (where each block sits), block classification (title, table, equation, signature), and inline confidence scores per page and per word. So when the model isn’t sure about a line in a scanned invoice,… Continue reading
-
OpenAI Codex Record and Replay: demo a Mac task once, the agent repeats it forever
OpenAI quietly turned Codex from a coding tool into a general computer-use agent. Record and Replay, shipped June 18 in the Codex macOS app (v26.616), lets you perform a workflow by hand — click here, fill that, hit submit — and Codex watches, then writes the whole thing up as a reusable skill. What it… Continue reading
-
gstack puts Garry Tan’s full Claude Code setup — 23 role-playing agents — in one install
Garry Tan, YC’s president, just open-sourced the exact Claude Code config he codes with. It’s called gstack, and it crossed 117K GitHub stars (+674 today) almost entirely on his name plus one irresistible pitch: clone a billionaire’s dev workflow with a single command. What it actually is gstack isn’t an app. It’s 23 opinionated slash… Continue reading
-
AI Berkshire (ai-berkshire): four investing legends, four agents, one verdict on your stock
Most “AI stock research” tools dump a wall of bullish-and-bearish hedging and call it analysis. AI Berkshire (ai-berkshire) refuses to. It’s an open-source agent framework that runs on Claude Code, and it forces a call: Pass, Conditional, or Grey Zone — with a concrete price range attached. The author claims a real-money +69% in 2024… Continue reading
-
Cognee replaces your graph DB, vector store, and session DB with one Postgres
Agent memory is having a moment, and Cognee just rode it to 23.7k GitHub stars — up 700+ in a single day. It’s an open-source long-term memory layer that gives AI agents persistent recall across sessions, instead of stuffing everything back into the context window. What it actually does Feed Cognee your data in any… Continue reading
-
DeepSeek DSpark pushes V4 inference up to 4× faster — and ships the training code too
DeepSeek didn’t release a new model this time. It made the existing one run faster. DSpark is a semi-parallel speculative decoding framework, and it’s already serving live traffic on DeepSeek-V4 Flash and Pro. HackerNews put it on the front page with 600+ points. What it actually does Speculative decoding drafts several tokens cheaply, then verifies… Continue reading
Top AI Product
Every day, hundreds of new AI tools launch across Product Hunt, Hacker News, and GitHub. We dig through the noise so you don't have to — surfacing only the ones worth your attention with honest, no-fluff reviews. Explore our latest picks, deep dives, and curated collections to find your next favorite AI tool.
Top AI Product 