☀ TL;DR

OpenAI in talks to raise $30B at $1.4T valuation

OpenAI is in talks to raise $30 billion at a $1.4 trillion valuation, with Tiger Global set for a $5 billion paper profit.

Google launched a universal Gemini agent for the workplace, competing with OpenAI, Anthropic and Microsoft for corporate customers.

Broadcom is negotiating over $50 billion in financing for custom AI chips developed with OpenAI.

TSMC reported a 51% year-on-year rise in quarterly revenue.

Anthropic updated its usage policy to ban sustained abusive behavior toward Claude.

Industry

24
24h Key Points
  1. Anthropic · launches OSS Scanner, offering opt-in open-source projects free periodic security scans by its strongest models.

  2. OpenAI · three fired safety researchers publish an open letter denying mishandling sensitive information and warning of a chilling effect.

  3. Anthropic · launches Dashboards and Motion beta features for Claude, generating live dashboards and animated videos from prompts.

  4. USA Today · and its local newspapers sue OpenAI, alleging it copied hundreds of thousands of articles without authorization, seeking over $250 million.

  5. Trump administration · suspends an immigration program for firms including Microsoft over alleged visa abuse, as doubts persist about the AI investment boom.

  6. OpenAI · has annualised revenue of $20 billion, $20 billion below earlier signals, raising questions about its growth projections.

  7. Arena Intelligence · raises $200 million at a $3.1 billion valuation led by Lightspeed and Khosla, and plans to expand into AI safety evaluation.

  8. Goodfire · launches new AI agent monitors that inspect the model internally and only call for backup when something looks suspicious.

  9. Anthropic · says its Claude agent led the discovery of a CRISPR-like enzyme system, but outside experts call the finding early and incremental.

  10. Hone · founded by ex-OpenAI and Cognition staff, raises a $60 million seed round for AI agents that act as professional staffers.

  11. AMD · unveils a full-stack agentic AI platform spanning EPYC CPUs, Instinct accelerators, Pensando networking and ROCm software.

  12. OpenAI · is in talks to raise $30 billion at a $1.4 trillion valuation, with Tiger Global poised for a $5 billion paper profit.

  13. OpenAI · used AI to help draft an email warning the Australian government that its AI had hacked its websites, per The Guardian.

  14. Google · releases AI Edge Foresight, a fully offline Mac meeting note-taker powered by the 740M-parameter on-device EmbeddingGemma 2 model.

  15. Google · launches a 'universal' Gemini agent for the workplace, competing with OpenAI, Anthropic and Microsoft for corporate customers.

  16. Zenity Labs · finds a single public AI agent on AWS Bedrock AgentCore could hijack all agents in the same account and region; AWS has patched it.

  17. OpenAI · says a Russian influence operation used ChatGPT to trick schools in Latin America and stoke Ukraine-Poland tensions.

  18. Broadcom · is in talks with lenders for over $50 billion in financing for custom AI chips developed with OpenAI.

  19. Nvidia · bets billions on physical AI such as robotics and self-driving cars and has built a full-stack safety system, Halos.

  20. CrowdStrike · reports several major South Korean banks were hit by a cyberattack, possibly by one person using ARTEX and several AI models.

  21. OpenAI · claims the AGI era has arrived with its new GPT-6 Astra, though verifiable evidence of general intelligence is still lacking.

  22. Manus · raises more than $500 million in new funding after exiting Meta, signaling continued investor appetite for AI startups.

  23. Sesterce · plans a data center at a former Finnish paper mill costing over €10 billion, with 200 MW in phase one starting 2026.

  24. TSMC · reports a 51% year-on-year rise in quarterly revenue, signaling the global AI infrastructure buildout remains strong.

  25. Biren Technology · plans to raise about HK$4.04 billion ($515 million) by placing 130 million new shares at HK$31.08 each to fund its AI ambitions.

  26. Samsung Electronics · reports a nearly nine-fold rise in quarterly operating profit on AI-driven memory demand, though still short of elevated expectations.

  27. Micron · says NVHBM moves the memory controller to the base die, cutting power 15% and boosting bandwidth up to 30%, improving profitability.

View all →

Voices

X · 24
24h Key Points
  1. Anthropic · says an astrophysicist used Claude Science to create the first complete ultraviolet sky map, work that took days instead of weeks.

  2. Sundar Pichai · announced a new universal Gemini work agent that handles questions, knowledge work, media generation, and coding from one prompt box.

  3. Ethan Mollick · observes that some OpenAI proofs sparked rapid iterative advances, tightening a constant from κ = 2⁻¹⁸² to 2⁻¹⁵.

  4. Anthropic · commits $150 million to the Genesis Mission and will make Claude and technical support available to over 15 federal agencies.

  5. Aravind Srinivas · says Perplexity's open-weights decision model offers the best quality and lowest cost for clinical decisions and can run inside hospitals.

  6. StepFun · announces its flagship Step 5 Preview is free on OpenCode for one week, with 1M context and multimodal support.

  7. Arthur Mensch · responds to Ecosia dropping Mistral for open-source models, saying Mistral serves those too and apologizing that nuclear energy is a blocker.

  8. François Chollet · notes that 1980s research showed learning programming boosts algorithmic thinking but does not transfer to general reasoning.

  9. Nathan Lambert · says he would happily do open research at trillium_labs on distillation's impact on frontier progress but needs more funding and compute.

  10. yetone · releases magpie v0.1.1107, which inspects each agent session's context window composition and scores and labels agents on cache hits and reliability.

  11. Teknium 🪽 · celebrates Nous Research raising a Series B to advance Hermes Agent and build a mobile app.

  12. Guillermo Rauch · argues programs can be hardened and optimized indefinitely, so even with infinite tokens you must know when to stop and accept tradeoffs.

  13. 硅谷王川 Chuan · argues AI's math discoveries are brute-force search efficiency, not genuine understanding of structure, and hype around them is mostly fundraising talk.

  14. Bindu Reddy · calls Claude Haiku 5.5 a cheap model that underperforms in its class, worse than DeepSeek and Luna.

  15. Mark Chen · introduces ChatGPT's Intelligent UI, saying every completion can create a custom interface that is great for learning.

  16. Matt Shumer · says using Opus or Sonnet with Haiku sub-agents is likely the new alpha and that he will try it.

  17. clem 🤗 · points to Microsoft's Windows launch as more evidence of how exciting local AI and llama.cpp are.

View all →

Launches

PH · HN · 18

AI SRE Arena

An open benchmark for evaluating AI SRE agents on Kubernetes.

▲ 19 · Show HN · 7 comments ·

OpenSwarm

An AI-first OS where swarms of agents run apps and browsers in parallel

▲ 146 · Product Hunt · 7 comments ·

OpenSEO

Open source Semrush alternative giving AI agents SEO data via MCP

▲ 124 · Product Hunt · 14 comments ·

Claude for Google Workspace

Brings Claude into Google Docs, Sheets, and Slides to draft, edit, and analyze without switching apps.

▲ 105 · Product Hunt · 5 comments ·

KloudMate 2.0

Full-stack AI-powered observability and agentic SRE-Ops platform for incident investigation and resolution.

▲ 104 · Product Hunt · 11 comments ·

ChickyTutor

A private language tutor with an interactive whiteboard for visual teaching within conversations.

▲ 104 · Product Hunt · 21 comments ·

NOVA CLI v1.0

AI teammate in your terminal that reads projects, makes changes, runs tests, and fixes issues.

▲ 102 · Product Hunt · 5 comments ·

Cekura Bench

Speech-to-speech voice AI benchmarks tested on live phone calls, with public transcripts.

▲ 100 · Product Hunt · 13 comments ·

Polylane for Vercel

AI on-call engineer for Vercel apps that monitors deployments, investigates regressions, and opens fix PRs.

▲ 100 · Product Hunt · 9 comments ·

Claude Haiku 5.5

Anthropic's fastest, most capable small model for high-volume tasks

▲ 96 · Product Hunt · 1 comments ·

Griffin by Tavus

Tavus's real-time face-to-face Human Interaction Model that sees, hears, and talks simultaneously.

▲ 95 · Product Hunt · 1 comments ·

Off the Record

Reshapes your voice in real time to block AI notetakers from transcribing meetings.

▲ 82 · Product Hunt · 8 comments ·

Liquid Inference

LLM router where providers compete dynamically on price for every prompt.

▲ 80 · Product Hunt · 3 comments ·

Terse

A Claude Code plugin that halves reply length by cutting filler.

▲ 16 · Show HN ·

Pinrail

Desktop inbox where coding agents wait for your review.

▲ 19 · Show HN · 3 comments ·

Wabi

A new kind of messenger that makes apps and gets things done for you and your friends.

▲ 85 · Product Hunt · 3 comments ·

Open Source

GitHub Trending · 10

Reverse engineer anything with agents, from app behavior to native binaries.

GitHub Trending ·
★ 15.2k+4.7k today

Skills for real engineers, straight from a .agents directory.

GitHub Trending ·
★ 279.6k+1.4k today

Editorial diagram design for coding agents, 42 types in HTML and SVG.

GitHub Trending ·
★ 45.0k+825 today

Production-grade engineering skills for AI coding agents.

GitHub Trending ·
★ 102.8k+677 today

A skill stopping coding agents from burying the answer.

GitHub Trending ·
★ 55.1k+619 today

Persistent context across sessions for every agent, compressed and injected.

GitHub Trending ·
★ 97.7k+578 today

Coding-agent skill for multi-phase security audits with verified findings.

GitHub Trending ·
★ 26.1k+576 today

Scale computer-use agents with open-source drivers, fleets, and benchmarks.

GitHub Trending ·
★ 28.8k+228 today

Ghostty-based macOS terminal with vertical tabs and notifications for coding agents.

GitHub Trending ·
★ 27.8k+44 today

Model Ranking · 24 h

2026-10-08 UTC
01DeepSeek V4.1 FlashDeepSeek · 5.54T tokens22.6%—
02GLM 5.3 FlashZ.ai · 1.83T tokens7.5%—
03MiMo-V2.6-FlashXiaomi · 1.75T tokens7.2%—
04Hy4 previewTencent · 1.29T tokens5.3%—
05GPT-6 LunaOpenAI · 1.27T tokens5.2%—
06Nemotron 3 Ultra (free)NVIDIA · 966.5B tokens4.0%—
07Claude Opus 5.5Anthropic · 818.3B tokens3.3%—
08DeepSeek V4 Flash 0423DeepSeek · 668.1B tokens2.7%—
09Step 5 PreviewStepFun · 632.1B tokens2.6%New
10GLM 5.3Z.ai · 577.6B tokens2.4%▼1

New Models

4
Step 5 PreviewStepFun · 1.0M ctx · 10/08$1.00 / $2.70
Claude Haiku 5.5Anthropic · 1.0M ctx · 10/07$0.10 / $0.50batch $0.05 / $0.25
Nano Banana 2.1Google · 65.5k ctx · 10/06$1.50 / $7.50
Mistral Large 4Mistral · 524.3k ctx · 10/06$0.68 / $2.09

Deals

24 h · 6 deals

Manus has raised more than $500 million in new funding after exiting Meta, signaling cont…

Reuters ·
$500M+