🧑‍💼 Z.ai Releases GLM-5

Good Morning! Chinese startup z.ai just dropped GLM-5, an open-source frontier model that sets a new record for low hallucinations, and it’s dramatically cheaper than top proprietary rivals. Plus, you’ll learn how to enable Claude Opus 4.6 for OpenClaw and Claude Code step by step.

Plus, in today’s AI newsletter:

• How to Enable Claude Opus 4.6 for OpenClaw and Claude Code

• Z.ai Releases GLM-5 With Record-Low Hallucinations

• DeepMind’s Aletheia Pushes AI Into Real Science

• AI Enters the Agentic Takeoff Phase

• 4 new AI tools worth trying

AI MODELS

🧑‍💼 Z.ai Releases GLM-5 With Record-Low Hallucinations

GLM-5 is the latest open model in z.ai’s GLM series, released under an MIT License and built for real-world, agentic knowledge work. It now leads the industry in knowledge reliability, outperforming Google, OpenAI, and Anthropic on hallucination benchmarks.

• Achieves a -1 score on the AA-Omniscience Index, the lowest hallucination rate ever recorded

• Uses a massive 744B-parameter MoE architecture with only 40B active per token for efficiency

• Introduces “slime,” a new async RL system that speeds up agentic training and long-horizon reasoning

• Priced at ~$0.80 input / ~$2.56 output per 1M tokens, ~6–10× cheaper than Claude Opus 4.6

GLM-5 shows that open models are no longer a compromise. With top-tier benchmarks, native agent workflows, and radically lower costs, it signals a shift toward open, execution-focused AI systems that can actually run an autonomous “office,” not just chat.

TOGETHER WITH HUBSPOT

Want to get the most out of ChatGPT?

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide.

AI MODELS

🧮 DeepMind’s Aletheia Pushes AI Into Real Science

Aletheia is an internal Gemini-based system designed for deep, scalable reasoning. It’s already solving open math problems, contributing to research papers, and tackling PhD-level challenges across disciplines.

• Scores up to 90% on IMO-ProofBench Advanced, a benchmark for rigorous math proofs

• Solved open problems, including four from the Erdős database

• Contributed to publishable research in algorithms, economics, ML optimization, and physics

• Acts as a reasoning partner: refining proofs, finding counterexamples, and linking ideas

Just 2.5 years ago, chatbots struggled with basic math. Now, AI is becoming a force multiplier for scientists, handling verification and deep reasoning so humans can focus on creativity, intuition, and big ideas. This isn’t automation of science, it’s collaboration.

AI NEWS

🚀 AI Enters the Agentic Takeoff Phase

Recent releases like GPT-5.3 Codex and Claude Opus 4.6 mark a clear shift: AI systems can now plan, execute, and complete complex projects with minimal supervision. Even more striking, they’re starting to accelerate their own development through feedback loops.

• New models can handle long, multi-step tasks independently

• AI is now contributing to improving future AI systems

• Task-length benchmarks show rapid growth in autonomous work

• Knowledge work disruption is expected as early as 2026

This is the transition from AI as a tool to AI as a teammate. The real advantage won’t go to those who debate the change, but to those who restructure work fastest around agentic systems. 2026 isn’t a prediction anymore, it’s a deadline.

HOW TO AI

🗂️ How to enable Claude Opus 4.6 for OpenClaw and Claude Code (full guide)

In this guide, you'll learn how to enable Claude Opus 4.6 in ClawdBot and ClaudeCode. You will also learn how to activate "Agent Teams" to build complex software in minutes and save money on API costs.

🧰 Who is This For

• Creators who want AI responses that sound more human and structured

• Developers who want cleaner reasoning, better code, and fewer hallucinations

• Power users building custom GPTs, agents, or system prompts

• Anyone curious how billion-dollar AI models are actually instructed

This gives you two superpowers:

1. Massive Memory: The AI can remember 1 million words (tokens) at once.

2. Agent Swarms: You can hire a "team" of AI agents to write code for you simultaneously.

Upgrading ClawdBot/OpenClaw

Currently, ClawdBot doesn't have a simple "Update" button for Opus 4.6. So we have to use a special "workaround" prompt to trick it into upgrading itself.

Step 1: Open a Fresh Chat

• Open ClawdBot.

• Click New Chat to ensure you are starting with a clean slate.

Now, we need to tell ClawdBot to research the new model and edit its own internal settings file to support it.

Copy this prompt exactly (by Alex Finn) and paste this into the chat box, hit enter and wait for Claude to confirm it has updated its configuration.

Step 2: Verify the Upgrade

Ask Claude: "Are you now running on Opus 4.6?"

It should confirm yes. You now have access to the 1 Million Token Context Window (it remembers way more) and 128k Output (it can write longer responses).

Step 3: Optimizing ClawdBot (Getting Better Results)

Now that you have the super-smart model, don't just ask it simple questions. Use "Reverse Prompting" to make it do the thinking for you.

Instead of telling it exactly what to do, we want it to tell us how to use its new power.

Copy and Paste this exact prompt: "Now that we are on Claude Opus 4.6 based on what you know about me and the workflows we've done in the past how can you take advantage of its new functionality to perform new workflows." Claude will list new capabilities tailored to you, such as:

"True Second Brain": Asking questions about conversations you had weeks ago.

"Overnight Tasks": Giving it a massive research project to run while you sleep.

Upgrading ClaudeCode & Enabling "Swarms"

This is the most powerful feature. "Agent Teams" allows Claude to spawn multiple AI employees to write code for you simultaneously.

Step 1: Open ClaudeCode

Open your terminal or command prompt where you run ClaudeCode.

Step 2: Enable the "Agent Teams" Feature

By default, this feature is hidden. You need to ask Claude to turn it on. Type the following command and hit Enter:

"I'm looking to enable agent teams which just became available in Claude Opus 4.6. Can you read the docs and help me figure out how to enable it?"

ClaudeCode will read the documentation and ask for permission to edit your settings. Type Yes and hit Enter.

Step 3: Reset ClaudeCode

Close your terminal and open it again to ensure the new settings are loaded.

🧑‍💻 Claude Opus 4.6: Built for long, serious work in coding and research

⚙️ GPT-5.3 Codex: Faster coding model with strong reasoning

🎨 Qwen-Image-2.0: Alibaba’s all-in-one image generation and editing model

🐦‍⬛ Raven-1: Tavus’ real-time emotion detection model for AI conversations

THAT’S IT FOR TODAY

Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!

See you tomorrow :)

- Dr. Alvaro Cintas