Highly recommend reading this guest post by theoretical physicist Matthew Schwartz to get a sense of how AI is helping science. He found Opus 4.5 to be roughly the level of a second-year grad student and it helped him accelerate his research by 10x.
LLMS
-
Wishing for the best AI model and harness to win
By
–
So many stories like this in Bali You book a beautiful luxury couple massage Beautiful temple with a jungle garden and nice staff welcoming you with a welcome drink Then you lie down and 15 minutes into the massage the trash burning starts, the entire room gets cooked and
-
Anthropic Bill: Model Fails at Instruction Following Task
By
–
He asked for $107 back on his Anthropic bill. Should have used a model that is better at instruction following…
-

Claude Gains Computer Control via Mouse Keyboard Screen
By
–
Anthropic is taking another big crack at computer use.
— The Rundown AI (@TheRundownAI) 23 mars 2026
Claude can now control your mouse, keyboard, and screen to use any app on your computer.
Can also be paired with the new Dispatch to remotely control everything from mobile: https://t.co/hAA6CqmFVCAnthropic is taking another big crack at computer use. Claude can now control your mouse, keyboard, and screen to use any app on your computer. Can also be paired with the new Dispatch to remotely control everything from mobile:
-

Claude MCP Tools Now Load On Demand Reducing Context Bloat
By
–
No more context bloat from unused MCP servers in your context. Claude, Cowork and Claude Code now load MCP tools on demand. Before and after:
→ View original post on X — @jiquanngiam, 2026-03-23 22:49 UTC
-

Anthropic Developing ‘Phone Use’ Agent Capabilities for Claude
By
–

BREAKING : Anthropic is also likely working on a Phone Use, so Claude will be able to make calls and execute tasks on your mobile device. Orbit Discovered by @M1Astra
-
LLM Comparison: Sovereign AI Deployment Strengths
By
–
Not all LLMs are created equal—each brings unique strengths to the table. 🚀
— SambaNova (@SambaNovaAI) 23 mars 2026
Thomas Vits from @InfercomAI breaks down what sets different models apart and how they power secure, sovereign AI deployments.
🎥 Watch the full session here: https://t.co/dRrVfeP40a
💡 Learn more… pic.twitter.com/PwcYIN47eyNot all LLMs are created equal—each brings unique strengths to the table. Thomas Vits from @InfercomAI breaks down what sets different models apart and how they power secure, sovereign AI deployments. Watch the full session here: https://
youtube.com/watch?v=nzsa1-
F3N2E&si=C0g5LBSdKDSj–00
… Learn more -
Claude Gains Autonomous Computer Use and Desktop Control Capabilities
By
–
BREAKING 🚨: Claude can now use your computer and apps to complete its tasks. The same also works with Claude Dispatch, where you can instruct Claude from your mobile device.
— 🚨 AI News | TestingCatalog (@testingcatalog) 23 mars 2026
A full-featured Computer Use 👀
"Research preview in Claude Cowork and Claude Code, macOS only." https://t.co/KQCWhF1kTE pic.twitter.com/4s5DesYUubBREAKING : Claude can now use your computer and apps to complete its tasks. The same also works with Claude Dispatch, where you can instruct Claude from your mobile device. A full-featured Computer Use "Research preview in Claude Cowork and Claude Code, macOS only."
-
People awakening to Claude models’ profound life-changing impact
By
–
People are slowly beginning to understand the profound changes models like Claude are bringing to their lives.
— Chubby♨️ (@kimmonismus) 23 mars 2026
And they're shocked. We'll be seeing many more clips like this. https://t.co/tCcwQldqIp pic.twitter.com/Go6PhxeDlrPeople are slowly beginning to understand the profound changes models like Claude are bringing to their lives. And they're shocked. We'll be seeing many more clips like this.
-

Anthropic’s Computer-Use Agent: Continuous Improvements and Daily Releases
By
–
Anthropic will have the best computer-use-agent. It just works and keeps getting better. No day without a new release.