Alibaba opened a preview of Qwen4-Max, the first flagship of its Qwen4 generation, with a 1M-token context and a focus on coding and agents. Agents kept spreading into the enterprise with Databricks' always-on Genie and Adobe's MCP server, while United Airlines banned AI agents from booking tickets. On the money side, OpenAI reportedly posted about $10B in Q3 revenue but a $10.9B operating loss, and Anthropic set an IPO price range of $70–78 per share.
From 10:00 on Nov 11 (Beijing time), Qwen4-Max-Preview, the first Qwen4 flagship, is available via Alibaba Cloud's Token Plan and the Qwen, QwenStudio and Qoder apps, with a 1M-token context, text/image/video input and a focus on coding and agents. The preview runs only until Nov 25, with no benchmarks, parameter count or open weights disclosed.
On Nov 11, Databricks launched second-generation Genie, which runs in the background on dedicated compute, splits work across parallel sub-agents and contacts people via a priority inbox only when a decision is needed; it adds Genie Ontology, Genie Search (100+ data sources) and Unity Gateway for per-agent spending caps and approval rules. MCP-based plugins for ChatGPT, Claude, Copilot and Grok put it in direct competition with Anthropic and Microsoft for enterprise agents.
On Nov 11, United amended its contract of carriage to bar AI agents from making, modifying or servicing tickets or buying services, citing data exposure, unintended purchases and conflicts with FAA/DOT rules. Delta and American have not followed; it is the latest example of businesses blocking shopping agents ahead of the holidays.
At Adobe MAX (Nov 10), Adobe turned its assistant into a proactive agent that gathers work context and released a local Model Context Protocol server letting ChatGPT, Claude and Gemini drive Photoshop, Illustrator, After Effects and Premiere. It opens one of the largest creative suites to outside agents.
Per The Information and the WSJ (summarized by Forbes on Nov 11), OpenAI's Q3 revenue rose nearly 50% quarter-on-quarter to about $10B, with enterprise revenue nearly doubling to roughly half the total; the adjusted operating loss was about $10.9B, or $1.09 lost per revenue dollar (down from $1.50). Figures are unaudited and come from investor documents.
Roadshow materials show Anthropic offering 825 million Class A shares (801.4 million from the company) at $70–78, about $64B at the top of the range, listing as ANT on Nasdaq; reports say seven strategic partners including Google and Nvidia committed to buy $26B of stock. The trading date is not yet confirmed.
Israeli startup Plurai raised a $13M seed round led by Team8 (Nov 11). It builds a digital twin of a customer's agent, runs simulated users to surface hallucinations and policy violations, then trains small models that can replace frontier LLMs and act as production guardrails.
Google DeepMind said its multi-LLM agent system AlphaProof Code generated about 6.5 million lines of Lean proofs verifying core security properties of Signal's PQXDH handshake and Triple Ratchet down to AArch64 machine code in about four weeks, with humans reviewing roughly 1% of results. The scope covers specified properties, not every possible attack; the paper and Lean code are public.
Google DeepMind (paper PDF) · GitHub google-deepmind/code-verification · Wired
On Nov 11, the Ministry of Science and Technology announced that Vietnam, in its first official entry in IMD's World Digital and AI Competitiveness Ranking (WDAIR 2026), placed 37th of 70 economies and 3rd of 6 assessed Southeast Asian economies. The result ties to Resolution 57's goal of putting Vietnam in the global top 50 for digital competitiveness by 2030.
If these digests help you make sense of technology, a small contribution keeps NHỊP picking out what is worth reading.
Scan the QR code with a Vietnamese banking app.
Download QR ↓ · View full image ↗