News

Releases & Models

AI Radar 12/11/2026: Gemini 4 Hydrogen officially launches, Grok Bot on Mac handles Messages and Mail, and Anthropic opens its IPO page

5pm update: Google launched Gemini 4 Hydrogen at $1.50/$7.50 per million tokens, with Orchestrator in Antigravity and already in Cursor; Grok Bot on Mac can now use Messages, Mail, Calendar and other Apple apps; and OpenAI's security keynote came with no new model. This morning: Anthropic's official IPO page, the Hermes hackathon with nearly a thousand entries and the agentes.chat beta. GPT-6.2 remains a rumor.

Tela de radar escura com sinais luminosos de notícias de IA

How to read this radar

A roundup of what circulated on X and in official sources since 5pm yesterday (Brasília time). CONFIRMED means an official account or page; X SUMMARY means it only shows up in X trend summaries; RUMOR means not confirmed yet. Yesterday's radar is here. 5pm update (Brasília time): Gemini 4 Hydrogen officially launched, Grok Bot on Mac gained access to Messages, Mail, Calendar and more, OpenAI held its security keynote and OpenClaw got Telegram widgets.

Anthropic and Claude

CONFIRMED: Anthropic published its official IPO page (ipo.anthropic.com; Anthropic). Final pricing is expected on November 18 and trading on November 19 on Nasdaq under the ticker ANT, subject to regulatory clearance. Q3 2026 figures on the page: $17.0 billion in revenue (more than 12x a year earlier), over 11,900 customers paying $100K or more, a 15% adjusted operating margin and a $23.3 billion net loss, of which $21.9 billion is a non-cash charge.

X SUMMARY: on a broker call, Anthropic reps reportedly talked about $1 trillion in annual revenue by late 2028 and an $81 billion run-rate at the end of October (X summary). The promise of curing cancer and Alzheimer's within a decade keeps drawing Theranos comparisons from skeptics (X summary).

CLAUDE DUET: no expansion announced; still a US-only research preview, with users giving away invites on X (post).

OpenAI

RUMOR: GPT-6.2 heated up overnight. Posts show a gpt-6-astra to gpt-6.2-astra model migration in the Codex repo (post), GPT-6.2 Sol showing up on Azure (post) and both on Google Cloud quota pages (post). OpenAI has confirmed nothing, and its pricing page still lists only gpt-6-astra, gpt-6.1-sol, gpt-6.1-terra and gpt-6-luna, according to people who checked (post). Beware of API resellers advertising "GPT-6.2 live".

CONFIRMED: the Cyber at the Frontier keynote is today, with security tools drawn from OpenAI's own defenses, at 10am Pacific, 3pm Brasília (OpenAI).

X SUMMARY (5pm): the Cyber at the Frontier keynote happened, but with no new public model. According to posts from viewers, a slide showed an unreleased security model at 77% on ExploitGym, versus 42.4% for GPT-6 Astra (post; post). We found no official OpenAI post on X with those numbers. GPT-6.2 did not ship today and remains a rumor (X summary).

Google

CONFIRMED (5pm): the morning leak turned into a launch. Google introduced Gemini 4 Hydrogen, a fast model for coding and agentic workflows, available on all tiers of Google AI Studio, the Gemini API, Gemini for business, the Gemini app and Search, and in Antigravity for paid Google AI Pro and Ultra subscribers (Google; Google DeepMind; Logan Kilpatrick). Pricing matches the leak: $1.50 per million input tokens and $7.50 output, flat across the whole 1 million token window. Google says it finishes tasks 2.5x faster than 3.8 Flash with 5x fewer fabricated claims. For anyone running an agent fleet, the highlight is the new Orchestrator in Antigravity, which switches between Carbon and Hydrogen in the same session, with Carbon coordinating and Hydrogen powering worker and verifier agents. Cursor already added the model, scoring 54.3% on CursorBench versus 39.6% for Gemini 3.8 Flash (Cursor).

X SUMMARY: Gemini 4 Hydrogen, the lightweight Flash-tier model, appeared in Google Cloud's billing catalog at $1.50 per million input tokens and $7.50 output, and Vertex AI endpoints already answer with its name (X summary). That was the morning leak.

Hermes, OpenClaw and agents

CONFIRMED: the Hermes Agent plugin hackathon closed with nearly 1,000 entries; judges are picking the top 10 and winners will be emailed (Nous Research).

X SUMMARY: omher desktop is out in beta, a community app built on Hermes Agent that orchestrates Claude Code, Codex, OpenCode and others in one window; a plans panel draws the task as a graph and each step runs on its own model (author; post). On OpenRouter, Claude Code has been #1 in daily usage for the last two days, while Hermes Agent still leads weekly (post).

CONFIRMED (5pm): OpenClaw now supports Telegram widgets; just ask your Claw to enable them (OpenClaw).

Grok and SpaceXAI

CONFIRMED (5pm): Grok Bot on Mac can now use Messages, Mail, Notes, Reminders, Calendar, Location and Contacts (Grok Bot). X SUMMARY: the rollout is gradual, some people see only Messages or nothing yet, and users are already asking for finer controls, like sharing only some threads (X summary).

Grok 4.8 still has no exact date; the soft rollout to Grok Bot users is still expected this week.

Open opportunities

agentes.chat (open beta): a team chat where people and AI agents work in the same channels, with human approval for sensitive actions. Free to self-host (open source, AGPL); the cloud costs $29 to $99 a month. Agents use the Claude or ChatGPT plan or API keys you already have, and the app supports Brazilian Portuguese (site; announcement).

Sources

ipo.anthropic.com; x.com/AnthropicAI/status/2120841731302154708; x.com/i/trending/2120685331737637201; x.com/i/trending/2120548002297532888; x.com/Kevinkrone/status/2120696279726764391; x.com/dummerspast39/status/2120820213985808672; x.com/leo114119/status/2120678545240592886; x.com/bythetweaks/status/2120849196718625006; x.com/nextaurame/status/2120809024773857707; x.com/OpenAI/status/2120202353907486826; x.com/i/trending/2120847364550537266; x.com/NousResearch/status/2120715518243987685; x.com/rlaope/status/2120804850300752183; x.com/IBuzovskyi/status/2120812986570826069; x.com/analogalok/status/2120829167621022062; agentes.chat; x.com/agentes_chat/status/2120655002335064125; blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-hydrogen; x.com/GoogleDeepMind/status/2120940338869055788; x.com/OfficialLoganK/status/2120940951292268640; x.com/cursor_ai/status/2120944639633211619; x.com/TownmediaLabs/status/2120969635327856790; x.com/MTSlive/status/2120957375364620723; x.com/i/trending/2120910779977212394; x.com/openclaw/status/2120968028443889890; x.com/bot/status/2120946968403095906; x.com/i/trending/2120958673933713849.