Google shipped four open-weight models under one name on April 2, 2026 — and the obvious choice, the biggest one, is probably wrong for most developers. With four Gemma 4 model sizes spanning edge devices to workstation-class reasoning, the question…
On March 7, 2026, Andrej Karpathy posted a GitHub link on X and a single sentence: “It’s basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code.” Two days later, the autoresearch…
Alibaba released Qwen3.5 on February 16, 2026 — Chinese New Year’s Eve — and the 397-billion parameter headline is designed to mislead you. Only 17 billion of those parameters activate per forward pass. The model trades punches with GPT-5.2 on…
Months of AI safety alignment training. Millions of dollars in reinforcement learning from human feedback. One sentence to reverse all of it. On February 5, 2026, Microsoft researchers published a paper demonstrating that a single training prompt — “Create a…
In the last week of January 2026, an open-source AI agent project cycled through three different names, survived account hijackings and crypto scams, and birthed a social network exclusively for AI bots. Welcome to the chaos that is OpenClaw—and the…
Moonshot AI just dropped Kimi K2.5—a 1 trillion parameter open-source model that does something none of its predecessors could: see, think, code, and coordinate a swarm of AI agents simultaneously. Released January 27, 2026, K2.5 isn’t just another model update.…
The Transformer architecture has ruled AI since 2017. But in January 2026, every major AI company—IBM, NVIDIA, AI21 Labs, TII—is shipping models built on something else. The pattern is identical: Mamba layers for efficiency, Transformer layers for reasoning, mixed architectures…
DeepSeek V3.2 is impressive on paper—685B total parameters, 37B active via Mixture-of-Experts (MoE), 128K context window. DeepSeek V3.2-Speciale scored 96% on AIME 2025, beating GPT-5 High’s 94.6% on mathematical reasoning. But if you want to run DeepSeek V3.2 locally, the…
A new paper from MIT CSAIL just dropped a paradigm shift for long-context AI. Recursive Language Models (RLMs) achieved 91.33% accuracy on extreme-length tasks where direct prompting scored 0%. Zero percent. That’s not incremental improvement—that’s a different approach entirely. Published…
