AI-Pulse

[ AI-Pulse ]

Your Friday Briefing

The news, papers and releases that matter.

Opus 4.8 ships with an orchestration brain

Opus 4.8 ships with an orchestration brain

Opus 4.8 lands paired with `/workflows` — a primitive that lets Claude design and run its own fleet of parallel…
Cheap models, big bills

Cheap models, big bills

OpenAI lost more than it earned last quarter. An AWS user got hit with a $30,000 surprise bill. Meanwhile, cheap…
Anthropic moves into the building

Anthropic moves into the building

Anthropic ships Claude inside Microsoft Office (GA), launches Claude for Small Business, and open-sources nine pre-built banking agents. Plus Codex…
Karpathy: vibe coding is over

Karpathy: vibe coding is over

Karpathy on the shift from vibe coding to agentic engineering. Anthropic-SpaceX compute deal doubles Claude Code rate limits. DeepSeek V4…
One Claude, 9 creative apps

One Claude, 9 creative apps

Opus 4.7 lapped the field on Vibe Code Benchmark (71%, up from sub-25% in 4.5 months). Plus Anthropic ships Claude…
The postmortem you'd want to read

The postmortem you'd want to read

Anthropic's April 23 postmortem — three stacked bugs explain seven weeks of weird Claude Code behavior. Same day, GPT-5.5 lands…
Opus 4.7 lands after a month of complaints

Opus 4.7 lands after a month of complaints

Opus 4.7 drops after weeks of community frustration, Anthropic's 2026 Agentic Coding Trends Report on the implementer→orchestrator shift, and OS-BLIND…
A model too powerful to sell

A model too powerful to sell

Claude Mythos finds 27-year-old security holes but Anthropic won't sell it. Plus Managed Agents in public beta, and the paper…
44 hidden flags inside Claude Code

44 hidden flags inside Claude Code

The Claude Code source map leak (44 hidden flags, an unreleased background agent, and a Tamagotchi pet system), plus the…
Claude can now use your computer

Claude can now use your computer

Anthropic's week — Claude Computer Use ships, Auto Mode and Auto Dream change how Claude Code works between sessions, and…
GPT-5.4 costs 3x Gemini for the same score

GPT-5.4 costs 3x Gemini for the same score

GPT-5.4 launch with 3x premium pricing, Anthropic's 80,508-interview qualitative AI study, and Google Stitch reshaping AI design tooling.
AI reviews your pull requests

AI reviews your pull requests

Claude Code Review lands on GitHub PRs, small fine-tuned models start beating frontier APIs, and Yann LeCun raises a billion…
AI agents are growing new eyes

AI agents are growing new eyes

The week AI agents got new eyes — LSP in Claude Code, NotebookLM cinematic videos, GitNexus, and 88-year-old Donald Knuth…
8 of 12 AI models went bankrupt

8 of 12 AI models went bankrupt

Qwen 3.5-35B running locally takes on the cloud, an ASIC chip hits 16,000 tokens per second, and 8 of 12…

[ NEWSLETTER ]

Stay Updated with Our Blog​

Subscribe to receive the latest insights and updates directly to your inbox.

By clicking Join Now, you agree to our Terms and Conditions.