In partnership with

Dear Readers,

What if a single open-source model could turn your ideas directly into functioning websites, presentations, or code agents—without detours or restrictions? That's exactly what GLM-4.5 from Z.ai promises: a hybrid agent model with thinking and acting capabilities that combines strategic reasoning, tool use, and creative coding in a single system. Best of all, it's open to everyone – licensed under MIT and Apache.

In this issue, we take a closer look at this game changer and what it means for the future of open AI. Also: An AI system that designs drugs from proteins, a blood test that reveals your biological organ age, and a pill that challenges Wegovy and Mounjaro. Plus, rumors about GPT-5, inspiring quotes, and a look at neglected areas of AI application. Lots of food for thought – stay tuned!


In Today’s Issue:

  • GLM-4.5 released: Zhipu AI’s open source model hits 3rd globally in benchmarks, supports agents, coding, and reasoning

  • Cursor 1.3 out: Adds terminal Agent collab, context tracking, and faster editing

  • Google drops MedGemma: Open medical models for imaging and note summarization

  • Gero’s ProtoBind-Diff: AI designs active drug molecules straight from proteins.

  • And more AI goodness…


All the best,

Top Article: GLM 4.5 released!

An exciting breakthrough: with GLM-4.5 and the leaner version "Air", Z.ai (formerly Zhipu AI) has released two open source models based on up to 355 billion parameters (32 billion active) and 106 billion parameters (12 billion active) respectively - and they bring reasoning, coding and agentic capabilities together in a hybrid approach.

In simple terms, the AI can both “think” strategically and respond immediately, help with programming and complete tasks independently using tools. In this way, it creates complex processes in dialog or code - all in one model. For example, it automatically generates entire PowerPoint slides, interactive mini-games or websites from a single input.

The added value for the AI community? GLM-4.5 achieves a top position in 12 standard benchmarks - 3rd place globally and number 1 among open source models.

The outlook: Imagine typing an idea and automatically getting an interactive demo website, a script agent, a complete slide package - all created by a single open source model. Will you be part of this future?

Why it matters: GLM-4.5 delivers top performance for reasoning, coding and agentics in an open platform - a real alternative to proprietary LLMs. For the first time, it enables developers to create cost-efficient, powerful agent systems with full transparency and freedom for customization.

The Takeaway
👉 GLM-4.5 and GLM-4.5-Air are MIT- and Apache-licensed - free use, fine-tuning, self-hosting possible.
👉 Agent-capable hybrid model with thinking/non-thinking mode and tool-calling: ideal for multi-level tasks and tool integration.
👉 Benchmark performance global top position among open source, cheaper than DeepSeek.
👉 Developer-friendly: compatible with vLLM, SGLang, OpenAI API interfaces, supports cutting-edge applications.

Sources:

Love Hacker News but don’t have the time to read it every day? Try TLDR’s free daily newsletter.

TLDR covers the best tech, startup, and coding stories in a quick email that takes 5 minutes to read.

No politics, sports, or weather (we promise).

In The News

Cursor 1.3 is out! 

You can now collaborate with Agent in your terminal, clearly see context window usage, and make faster edits.

EU AI Act: Concerns Regarding Innovation and Copyright

xAI supports AI safety and will sign the EU AI Act’s Code of Practice on Safety and Security. Although it advocates for AI safety, some parts stifle innovation and have overly broad copyright provisions.

Graph of the Day

Gero's ProtoBind-Diff: AI decodes nature's hidden library of active compounds

A new AI model called ProtoBind-Diff generates novel, biologically active molecules directly from protein sequences – without the need for complex structural models. Trained on over a million protein-ligand data sets, it opens up fast paths to target-specific drugs. For longevity research, this means discovering regulatory-ready molecules faster and precisely targeting potentially complex aging mechanisms.

Blood test determines organ age and predicts disease risks

A groundbreaking study in Nature Medicine described a machine learning-based blood protein panel that estimates how old individual organs appear biologically. A higher organ age correlates clearly with a higher risk of disease and a shorter life expectancy—conversely, youthful brain and immune values signal a longer, healthier life. Such individualized markers can radically improve preventive strategies.

Orforglipron

A brilliant breakthrough: Orforglipron is the first daily oral GLP-1 pill to show in Phase III that it is as effective for weight loss as injectable options. In studies, patients lost an average of around 8% of their body weight with comparable safety but more convenient dosing. A potentially affordable, user-friendly alternative to Wegovy and Mounjaro with approvals expected by the end of 2026.

Put your AI-driven Longevity Breakthrough in front of 200,000+ People

Working at the intersection of AI and human longevity? We’re inviting researchers and innovators to submit their work to be featured in Superintelligence, the leading AI newsletter with 200k+ readers.

Whether you’re using AI to extend lifespan, improve healthspan, or accelerate biomedical breakthroughs, email your arXiv.org link or a summary to [email protected] with the subject line “Longevity Submission”. If selected, we’ll contact you for a potential feature.

Question of the Day

Do you regularly use Open Source models?

Login or Subscribe to participate

Tweet of the Day

Rumors, Leaks, and Dustups

According to the latest leaks, GPT-5 will be better in coding than Claude 4.

In addition, GPT-5 has already been found in the MacOS app!

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

  1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

  2. Master AI tools, tutorials, and news in just 3 minutes a day

  3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

How'd We Do?

Please let us know what you think! Feel free to reply to this email with suggestions (we read everything)!

Login or Subscribe to participate

Reply

or to participate

Keep Reading

No posts found