Anthropic Just Doubled Claude's Usage Limits for Everyone

  Welcome back! Anthropic just doubled Claude's usage limits in a surprise weekend move, Elon Musk is admitting xAI needs to be rebuilt from scratch, Meta is reportedly preparing to cut 20% of its workforce to fund its AI bet, and a pet owner with no science background used four AI models to design a cancer vaccine that actually shrank his dog's tumor.


Anthropic Doubles Claude's Usage Limits Across Every Plan
Article content

Anthropic dropped a surprise weekend announcement: double usage limits across all Claude plans for the next two weeks. The boost applies automatically outside peak hours, with no action needed. The company also made its full 1M-token context window generally available for Opus 4.6 and Sonnet 4.6 at no additional cost.

The details:

    Double Usage: 2x limits on Free, Pro, Max and Team plans for two weeks, applied automatically outside peak hours (weekdays 5am to 11am PST). 
    1M Context Window: Full 1M-token context now GA for Claude Opus 4.6 and Sonnet 4.6, enabling uploads of entire codebases and long-form documents. 
    No Price Increase: Both the usage expansion and the context window upgrade come at no extra cost. 
    All Tools Included: The promotion covers all Claude tools, not just the chat interface.

The 1M-token context window is the bigger move here. For developers and researchers working with large codebases or dense documentation, this changes the scope of what a single conversation with Claude can handle.


Musk Admits xAI "Was Not Built Right" As Rebuild Begins
Article content

Elon Musk publicly stated that xAI "was not built right" and needs a ground-up rebuild. Nine of the original 11 co-founders have now departed, leaving only Manuel Kroiss and Ross Nordeen. The admission comes as SpaceX, which owns xAI, prepares for a public listing later this year.

The rebuild:

    Latest Departures: Zihang Dai and Guodong Zhang left the company, with Zhang reportedly blamed by Musk for Grok's coding shortfalls before exiting. 
    New Hires: xAI brought in senior Cursor engineers Andrew Milich and Jason Ginsberg, both reporting directly to Musk, to accelerate Grok's coding capabilities. 
    Coding Deficit: Musk has publicly acknowledged that Grok is "currently behind" on coding versus frontier competitors. 
    IPO Pressure: The full rebuild is happening while SpaceX prepares for one of the largest public listings in recent history.

Rebuilding from scratch while preparing for an IPO is an extraordinary challenge. Musk is betting that new talent and a clean architecture can close the gap with OpenAI and Anthropic, but the timeline to prove that is tightening.


Meta Reportedly Weighing Layoffs Affecting 20% Of Its Workforce
Article content

Meta is reportedly considering layoffs that could cut 20% or more of its nearly 79,000 employees. The move would offset aggressive AI infrastructure spending, including $600 billion earmarked for data centres by 2028.

The pressure points:

    Potential Scale: A 20% cut would impact roughly 16,000 people across the company. 
    AI Investment: Meta has committed $600B to data centre buildout, one of the largest infrastructure bets in tech history. 
    Recent Acquisitions: The company's purchase of Manus adds to its AI portfolio and its cost base. 
    Official Position: A Meta spokesperson described the reporting as "speculative reporting about theoretical approaches."

Meta is making one of the biggest infrastructure bets in corporate history while potentially cutting the workforce needed to operate it. A 20% reduction is not efficiency trimming. It is a structural bet that AI can absorb the work.


AI-Designed Vaccine Shrinks Rescue Dog's Tumor
Article content

Sydney AI consultant Paul Conyngham built a custom mRNA cancer vaccine for his dog Rosie by chaining ChatGPT, Grok, DeepMind's AlphaFold and a university genomics lab. One tumor shrank by half after the first injection.

How it worked:

    Diagnosis: Rosie was diagnosed with mast cell cancer in 2024 and given months to live after chemo and surgery failed. 
    AI Pipeline: Conyngham used ChatGPT to map the research, paid $3,000 for genomic sequencing and ran the data through AlphaFold to model mutations. 
    Vaccine Design: The UNSW RNA Institute helped produce the vaccine, with the final construct designed using Grok. 
    Results: One tumor shrank 50% after a December injection. A second vaccine targeting non-responding tumors is now in development.

Try this yourself:

Conyngham's approach demonstrates what is possible when you chain multiple AI models together for a single complex problem. The tools he used (ChatGPT, Grok, AlphaFold) are all publicly accessible. The breakthrough was not any single model, but the pipeline connecting them.