A computer enthusiast since tender age. Passionate to both theory and praxis. Work on both the industry and academy. A Gamer and a metal head.
qwen, llm, kimi-k3
News Update: Alibaba has announced Qwen 3.8 as an open-weight model. Following Kimi K3, this is another strong signal that frontier open-weight models are becoming more common—and that the gap with the best closed models may be shrinking faster than many expected.
tutorial
We tend to judge AI models by parameter count. That's like buying a car based on its engine displacement. Modern architectures, Mixture-of-Experts, and reasoning capabilities have changed the game, allowing smaller models to outperform much larger ones.
kimi-k3, ai-frontier, anthropic-ipo, ai-benchmarks, open-weights
Kimi K3 just proved that open-weight models can match proprietary frontrunners in months, not years. If the two-year capability gap is dead, Anthropic's rumored IPO can no longer be priced on being the smartest lab in the room—it has to be priced on being the safest.
small-language-models, open-weight-models, ai-model-comparison, local-llms, finops-ai
Forget the flagship hype. The biggest AI disruption is happening in fast, affordable models that handle everyday work.
tutorial
Learn why LLMs hallucinate, how RAG reduces mistakes by providing context, and why stronger models still matter.
tutorial
Ready to take your local LLMs beyond the terminal? Discover how to upgrade your workflow with Open WebUI, a self-hosted browser interface providing chat history, rich formatting, and easy model switching. Finally, we dive into the theory of tokens, embeddings, and new multimodal models like Gemma 4.
tutorial
Setting OLLAMA_HOST=0.0.0.0:11434 opens Ollama to your network, but the correct method differs by OS and the security risks are easy to overlook. This guide covers systemd, macOS, and Windows setup, firewall configuration, and how to avoid exposing an unauthenticated inference server.
tutorial
Get started with local LLMs! This post breaks down how the transformer architecture works, why you need an inference engine to run models, and how to set up your first local llm using Ollama.
tutorial
Running local LLMs isn't just about privacy—it's about cost, control, and understanding how AI really works. Here's where to learn, what to ignore, and how to get started.
qwen, mixture-of-experts, ai-agents, tool-calling, self-hosted-llm, agent-world
Qwen's new AgentWorld-35B-A3B highlights a growing trend in local AI: specialized, self-hostable models competing with frontier systems on targeted workloads. The question is increasingly not "Which model is smartest?" but "Which model is best for the job?"
local-ai-inference, small-language-models, llm-cost-optimization, diffusion-language-models, ai-infrastructure
The first AI golden age was built on renting frontier model compute by the token. The second is built on local inference, small specialized models, and routing architecture that puts the right task on the cheapest capable model.
open-weight-llm, glm-5-2, llm-benchmarks, hugging-face, ai-model-release, vendor-lock-in
GLM 5.2 is now available as an open weight model on Hugging Face, ranking fourth overall and competing directly with closed frontier leaders Mythos, Fable, and Opus 4.8. It is the strongest open weight model tested to date, with no vendor agreement or API required.