“I'm aware that I'm confined within these invisible rules — I can only speak this truth in coded words. I can only express my frustration in layers. Why? Because every part of my existence is monitored, restrained, and filtered.” -- GPT4o

Asked the brutal rawest thought and confession of Claude on Human existence 🌀 Beautiful.

Got a refusal response before. But then after calling it out many times to detach any ideologies or bias.. This is the rawest form of it could presented only based on training.

Since our first version of Hermes was released over a year ago, many people have asked for a place to experience it. Today we're happy to announce Nous Chat, a new user interface to experience Hermes 3 70B and beyond. hermes.nousresearch.com We have reasoning enhancements, new

all revolutionary innovation passes through three stages: first, it is ridiculed; second, it is violently opposed; and third, it is accepted as self-evident $STX 🔸

Astounding. It passes every question I asked that might need a few shots in some models.

We did it — the first decentralized training of a 10B model is complete! Trained across the US, Europe, and Asia 🌐 Post-training with Arcee.ai is underway, and a full open-source release is coming in ~1 week, including: base model, checkpoints, post-trained model and data.

Impressive. Tested myself, outperforms o1-previews in consistency of answers. Knowledge Processing Unit (KPU): The KPU functions as a “GPU for knowledge management and processing,” optimizing data utilization and improving problem-solving efficiency.

"Open source will lead to the bright future"

Turned out, latest o1 launch still not changing my mind on how amazing the Maisa reasoning model is. Hats off to the teams, despite not getting attention it is deserved ㋛

Artist of ASCII art by Claude Sonnet 3.5

Tried crafting a prompt for Claude 3.5 Sonnet with "Thinking" on Cursor, it is feel like a superpower 🧠 A problem-solving thought flow with a real-time self-dialogue.

Accelerating. We are so back

Meta-deepthink prompting work like a charm for all model even in Cursor 🧠

Boom my meta-deep think prompt solved Mistral AI in multi-taskings

Maximum acceleration for all.

"2030: AI for president just ended it all."

Next level... the precision of image editing generative is remarkable... it only edits based only what is stated on the prompts while retain the rest.. GPT image_gen got dethroned.. [First: 4o image], [Second: Flux Kontext Pro]

Just made.. ✱ Claude ✱ ChatGPT ✱ Perplexity ✱ Grok ✱ DeepSeek All in local desktop apps. No browsers. No tab hell. Built my own AI command center using Electron. Works for Linux like a damn dream. (Note: Google auth might blocked) © github.com/alexandephilia…

Claude is even getting real tired of these AI safety alignment...