Moonshot AI Launches Kimi K3 — World’s Largest Open-Weight Model CNBC / Bloomberg / TechCrunch

Chinese startup Moonshot AI launched Kimi K3 on July 16, a 2.8-trillion-parameter mixture-of-experts model that immediately claimed the #1 spot on Arena.ai’s Frontend Code Arena with a 76% pairwise win rate against Claude Fable 5. Shipped in two variants — K3 Max for chat and agents, K3 Swarm Max for large-scale parallel processing — the model features a 1-million-token context window, native multimodal understanding, and beats OpenAI GPT 5.5 and Claude Opus 4.8 on coding and agent benchmarks. Full open weights are due July 27; pricing starts at $3/M input tokens.

🤖 Frontier Models

Kimi K3: Moonshot AI’s 2.8T Open-Weight Model Tops Coding Arena — CNBC

Moonshot AI’s Kimi K3 is the largest open-weight model released to date, arriving with a 1M-token context window and two purpose-built variants. The model outperforms OpenAI’s GPT 5.5 and Anthropic’s Claude Opus 4.8 on multiple benchmarks while still trailing Fable 5 and GPT 5.6 Sol overall — a significant narrowing of the US-China frontier gap. The TechCrunch framing of Kimi as “threat or menace” reflects concern in some quarters about the model’s open-weight release enabling downstream fine-tuning with minimal guardrails.

Google Delays Gemini 3.5 Pro After Internal Testing Disappoints — TechCrunch The flagship model has been pushed back by several months after falling short on coding and complex long-horizon reasoning; Alphabet shares fell roughly 4% on the news.

Claude Fable 5 Free Access and Elevated Rate Limits Expire Today — Anthropic The Fable 5 free tier on paid plans and the 50%-higher Claude Code weekly rate limits both end at 11:59 PM PT tonight.

🌍 Policy & Geopolitics

China Launches WAICO: A 29-Nation AI Cooperation Alliance — Al Jazeera

At the 2026 World AI Conference in Shanghai on July 17, President Xi Jinping announced the World Artificial Intelligence Cooperation Organisation (WAICO), an intergovernmental body with 29 founding nations spanning the Global South — including Brazil, Indonesia, South Africa, Russia, and Pakistan. Xi pledged 5,000 AI training slots for developing nations and framed open-source AI as a public good, positioning the initiative as a direct counter to US export controls. UN Secretary-General António Guterres attended the launch.

UN Urges Binding AI Governance Amid ‘Catastrophic Harm’ Warnings — UN News A global coalition at the UN called for binding AI safety frameworks, warning that uncoordinated frontier model development poses catastrophic societal risks.

💼 Enterprise & Products

Anthropic Revenue Run-Rate Tops $30 Billion; Enterprise Customers Double — Anthropic

Anthropic’s annualized revenue has surpassed $30 billion — up from roughly $9 billion at end-2025 — driven by explosive enterprise adoption. The count of business customers spending over $1 million annually has more than doubled to over 1,000 since the company’s Series G announcement in February, a doubling in under two months. Anthropic also announced expanded partnerships with Google and Broadcom for multiple gigawatts of next-generation compute.

Claude for Teachers Gives US K-12 Educators Free Premium Access — Anthropic Launched July 14, the program gives verified US K-12 teachers one free year of premium Claude with standards-aligned lesson planning, FERPA-compliant student data protections, and math tools via ASSISTments. A pilot in Detroit Public Schools will study effects on educator well-being; critics have raised questions about AI influence in classrooms.

Microsoft Project Perception: A Multi-Model AI Security Tool — Tech Startups Microsoft’s forthcoming cybersecurity product dynamically routes tasks to the best-fit model from Microsoft, OpenAI, or Anthropic — positioning the suite as a cheaper, globally distributed alternative to single-vendor security AI.

🎨 AI & Culture

Dave Eggers Told OpenAI Staff ChatGPT Was “Silencing an Entire Generation” — The Verge

Author Dave Eggers — invited by Sam Altman to address roughly 200 OpenAI employees — used the platform to deliver a blunt critique: generative AI is preventing a generation from developing their own creative voice. The talk was reportedly received poorly by staff, making it an unusually candid window into the tension between AI insiders and the literary world.

‘Odyssey’ Director Christopher Nolan Calls AI an “Obvious Trojan Horse” — TechCrunch “Everybody knows the Greeks are inside” — Nolan invoked the myth to warn that AI’s surface-level benefits mask risks that society is choosing not to examine.

I Hate That I Don’t Hate This Suno Song — The Verge

🛠️ Developer Tools

Nonprofit Current AI Is Building the “World Wide Web of AI,” Free for All — TechCrunch Current AI aims to build open, interoperable AI infrastructure accessible across cultures and devices — framing its mission as creating an internet-equivalent layer for AI that no single company controls.

Claude Code 2.1.215: /verify and /code-review Now Manual-Only — Claude Code Changelog

Generated by claude-sonnet-4-6 · 2026-07-19T10:00:00Z