AI Pulse

What's moving in AI

Official lab announcements, top model rankings from Chatbot Arena and OpenRouter, and the day's most important AI news — curated and updated daily.

Last updated: Apr 8, 2026 · Back to home

From the labs

Announcements

Official posts from AI labs and key researchers, embedded directly.

Introducing Project Glasswing — powered by Claude Mythos Preview, Anthropic's most capable model, an urgent initiative to secure the world's most critical software.

Loading tweet...

Meta launches Muse Spark — a multimodal creative AI model generating images, audio, and video from a single prompt.

Loading tweet...

Gemma 4 launch — Google's most intelligent open models, built with Gemini 3 technology, now available for local hardware with Apache 2.0 licensing.

Loading tweet...

Anthropic raises $30B at a $380B post-money valuation to deepen research and power infrastructure expansion.

Loading tweet...

Rankings

Top models

Ranked by Overall Elo — crowdsourced head-to-head votes from Chatbot Arena. Pricing from OpenRouter. New models flagged.

#Model
1
2
3
4
5
6
7
8
9

Today's signal

Latest news

Curated from today's digest — top stories with editorial context.

NewsThe Verge

Anthropic unveiled a frontier model it says is too dangerous to release publicly

Claude Mythos Preview, a model above Opus, reportedly found thousands of severe OS and browser vulnerabilities. Anthropic restricted it to Project Glasswing, a defensive-security initiative with major tech partners — a clear signal that frontier capability is becoming a controlled strategic asset.

Read
NewsOpenAI

OpenAI published a policy memo on the social contract of the intelligence age

OpenAI's Industrial Policy for the Intelligence Age proposes public wealth funds, robot taxes, reduced workweeks, and broad AI access as automation accelerates. A frontier lab is no longer just proposing products — it's drafting the economic logic for a superintelligence transition.

Read
NewsArs Technica

Google launches Gemma 4 under Apache 2.0 — open models get more strategically usable

Gemma 4 adds four local-first open-weight variants, better agent support, native function calling, structured JSON output, and a switch to Apache 2.0 licensing. In a market where closed-model access can tighten overnight, more permissive open models become operationally — not just ideologically — attractive.

Read
NewsTechCrunch

OpenAI raises $122B at $852B valuation — the largest private funding round in history

Amazon committed $50B, NVIDIA invested $30B. OpenAI is now generating $2B in revenue per month with 900M weekly active ChatGPT users. A superapp merging ChatGPT, Codex, and the Atlas browser is in the works.

Read
NewsThe Verge

Anthropic reaches $30B ARR — capability control looks different at this scale

Anthropic's ARR jump frames every safety decision — including the Mythos/Glasswing restriction — as also a strategic market move. Once a lab has both frontier models and serious revenue gravity, these things can't be separated.

Read

Read deeper analysis on the blog →