ThursdAI - The top AI news from the past week

ThursdAI - Opus 1M, Jensen declares OpenClaw as the new Linux, GPT 5.4 Mini & Nano, Minimax 2.7, Composer 2 & more AI news

1h 32min

This ThursdAI episode covered major AI advancements, including Anthropic making Opus 4.6's 1 million token context window default and OpenAI's release of cheaper, code-optimized GPT 5.4 Mini and Nano models. A key highlight was NVIDIA CEO Jensen Huang declaring OpenClaw as the "new Linux," emphasizing the critical need for an agent strategy for every company. The episode also discussed new competitive models like Minimax M2.7 and Cursor Composer 2, alongside insights into model reasoning and new voice technologies.

Summarized by Podsumo

🎧 Listen 🎙️ Ask about this episode

✨ Key Takeaways

1

Anthropic Opus 4.6 now defaults to 1 million token context window

, making long conversations and complex tasks more efficient and affordable, leveraging its strong performance on long-context benchmarks.
2

NVIDIA CEO Jensen Huang proclaimed OpenClaw as the "new Linux,"

asserting that every company needs an OpenClaw strategy, and introduced NemoClaw as an enterprise-hardened version addressing security concerns.
3

OpenAI launched GPT 5.4 Mini and Nano models

, optimized for coding and sub-agent tasks, offering significantly cheaper and faster performance, nearly matching the main model's OS World scores.
4

Minimax M2.7 was introduced as a "self-evolving model,"

achieving near-SOTA performance (56% on Swibbench Pro) through over a hundred autonomous optimization loops, with zero lines of human code.
5

Wolfbench analysis revealed that while GPT 5.4 benefits from increased reasoning, Claude Opus 4.6 can "overthink"

in agentic tasks, leading to worse performance and higher costs with maximum reasoning settings, highlighting the importance of balancing thought and action.

💬 Notable Quotes

"Every company needs an OpenClaw strategy, and then every SaaS company will become a GAS company, a GENTIC as a service company."

— Jensen
"The agentic revolution absolutely exploded the number of tokens that a person can like realistically use."

— Alex Volkov
"It's just insane. 50 cents per million tokens for input two dollars and 50 cents per million tokens on output. That's even more affordable than 4.6 on it."

— LDJ