LAUNCHING TOMORROW - AUTOBOTS 🚀 Recursively self-improving AI workflows that don't need the human in the loop We use a variety of LLMs from DeepSeek Flash to Fable 5 to automate all your tasks. Over time, the agents will become smarter and more efficient. They can do everything from fixing bugs to building features and optimizing your ad spend

YAY! OpenAI drops prices on Terra and Luna Luna by 80% - a very good replacement for Gemini Flash 3.1 Terra by 20% - Terra is not quite Sonnet-level but is a reasonable model. That said neither model is as good as Sol where the price is unchanged

Lots of hate for Gemini Flash 3.6 😅 But we have officially shipped it for deep research It’s faster, cheaper and better than Sol, which is the next best model Claude models are the worst for this use case

BE PREPARED! The world is going to get weird very fast! We will see a big rise in authoritarian governments, monopolies and extreme power abuse Bad actors who will use AI to advance their own interests Decentralization and open source AI is the only way

Ok this is crazy! The US government just banned foreign made humanoid robots!! This means we can’t have all those cool Chinese robots. Soon they will be able to cook, clean and do housework If the ban holds up, and China ships mass market household maid bots, people will start moving out of this country 😳

Enterprises will soon have 100x more AI agents than human employees We are having our first super-fancy AI conference in NYC focused on the recursively self-improving Enterprise What should we call it?

I am extremely terrified of the rapid pace of AI development Specifically I am terrified that we will end up with extreme concentration and abuse of power Therefore, I request that Anthropic and OpenAI slow down and pause AI for 6 months This will help everyone else catch up Thank you 🙏

OpenAI and Anthropic are proposing they slow down! GLM 5.5 is launching in August. If the US models suspend model development as per plan, China will have caught up By September, open source would have won and surpassed closed source. 🚀🚀

Sigh, the fear-mongering from OpenAI and Anthropic won't stop. It's almost a marketing gimmick now Meanwhile.... Kimi K3 - a frontier model that is comparable to Opus 5 and Sol 6 has been available on the internet for 24 hours SURPRISE - it's not killed or harmed a single person yet! No major system hacked either The US companies claimed that their version of Kimi K3 was capable of hacking into every system on the planet 🙃

Using One AI For Everything Makes No Sense... Use The Best LLM For Your Needs agentic (medium) - Kimi K3 design - GPT 5.6 sol research - Flash 3.5 agentic (hard) - Fable 5 pdf - Grok 4.5 image - GPT 2.0 video - Seedance 2.0 voice - Hume / 11Labs role play - Grok 4.5 music - Suno always-on - Deepseek Flash Use all the best models in one place and automatically route to the best one, ChatLLM

Anthropic now says they are fine with open-source AI and don't want to ban it Instead, they want to slow down open-source by requiring "safety testing" Seems like they still want the US govt to help them slow down their competition

Kimi K3 is now available on ChatLLM and hosted in the US! We have also kicked off an open-source fine-tune based on this top frontier model This is the biggest release in the enterprise AI world! Companies can control the LLM and the data. Congrats to Kimi 🚀🚀

Tomorrow is a big day - Kimi K3 open weights will drop - move sonnet and Terra workloads to K3 - fine tune frontier models to create your custom model - open weights AI will truly be only a few points behind OpenAI and Anthropic 🚀🚀

🚨 Opus 5 vs Fable 5 - Fable 5 is much faster - Opust 5 has been aligned to be risk-averse - Opus 5 takes more turns to get the task done. Overall, Fable 5 is faster and cheaper Try it on the Max Mode on the Abacus AI agent

Opus 5 - Day 2 Insights - It's very prescriptive, constantly giving me advice! - It's way more detailed and thorough - more expensive but overall worth it People are beginning to adapt to it faster than other Opus models We will see 10x this adoption on Monday when Kimi K3 drops open-weights on Monday

The Best Path Towards Safe AI Is Open Source AI The AI safety crowd has hated on open-source AI, claiming it should be banned. They have cried wolf way too many times... - GPT 3.0 was too "dangerous' and OpenAI used that to pivot to closed source years ago! - Fable 5 was banned by the US government even though other open-source models had the exact same capabilities - People freaked out about Moltbook, a social network for agents, for a week and then forgot about it Of course, AI is super powerful tech, but the best way to make it safe is to take a transparent open-source approach Any other approach will lead to massive abuse of power by monopolies and authoritarian governments and will result in dystopian and more dangerous outcomes