Kimi K2, GLM-5 and DeepSeek-V4 all trained with a Newton-Schulz-orthogonalized optimizer that Moonshot…
Kimi K2, GLM-5 and DeepSeek-V4 all trained with a Newton-Schulz-orthogonalized optimizer that Moonshot AI measured at roughly 2x AdamW's compute efficiency.
It was the morning that my resignation letter was read 150 million times. In the same week
It was the morning that my resignation letter was read 150 million times. In the same week, a model with a third of the unit price became an official version, and the photos are starting to come with proof of authenticity. The quickening side and the checking side moved in the same week. 1️⃣ Unit price 1/3, deadline is September 14th 2️⃣ Gemini resides on Windows 3️⃣ Prove it's not AI 4️⃣ The person inside got off.
I compared DeepSeek-v4.1-Flash against Opus 5 and GPT-5.6 Sol.
Same prompt with the /design command from Command Code. Opus 5 is extremely good at UI/UX, but DSV4.1-Flash was almost 90% cheaper than GPT-5.6 Sol and Opus 5💀 What do you think? is DSV4.1-Flash really that good?
Get more AI agent work done without paying for every API call.
I connected DeepSeek V4.1 Flash to Hermes using Token Harbor. Then used it for: • Agent workflows • Games • Websites • Productivity tools For now, the API route is free. Want the SOP? DM me.
You have reached the end of the archive
All of deepseek