https://blog.mattiaperetti.it/blog Backend architecture, keeping software running in production, and building AI agents. en-us undefined (Mattia Peretti) undefined (Mattia Peretti) Sun, 12 Jul 2026 00:00:00 GMT https://blog.mattiaperetti.it/blog/choose-the-right-ai-model-for-the-job https://blog.mattiaperetti.it/blog/choose-the-right-ai-model-for-the-job Model choice is an engineering decision that shows up in both output quality and your bill. This walks through models.dev and artificialanalysis.ai as the two resources for shortlisting on specs/price and then checking capability, cost per task, and latency, with worked comparisons across GPT-5.6 Sol, Claude Fable 5, Gemini, LongCat-2.0, and Seedance 2.0. Sun, 12 Jul 2026 00:00:00 GMT undefined (Mattia Peretti) aillmagentic-aitokens https://blog.mattiaperetti.it/blog/local-llm-coding-agent-setup https://blog.mattiaperetti.it/blog/local-llm-coding-agent-setup I moved my coding agents off hosted APIs and onto two machines I own, a MacBook Pro M5 Pro as the client and a gaming PC with an RX 7800 XT as the model server. Model capability was rarely the blocker. The blockers were unified memory limits, inference engines that mangle tool call formats, a model stuck in a loop, and a laptop that kept freezing for reasons I still have not pinned down. Wed, 23 Sep 2026 00:00:00 GMT undefined (Mattia Peretti) aillmlocal-llmcoding-agentsself-hosted https://blog.mattiaperetti.it/blog/local-llm-tipping-point https://blog.mattiaperetti.it/blog/local-llm-tipping-point Colibrì, a ~1,300-line pure C inference engine with zero dependencies, runs the 744B-parameter GLM 5.2 MoE model on a laptop with 25 GB of RAM and no GPU by streaming int4-quantized experts from NVMe. Cold-cache throughput is 0.05 to 0.1 tokens per second, so speed is the honest limiter, but it is an existence proof that frontier-scale local inference works on commodity hardware. Fri, 17 Jul 2026 00:00:00 GMT undefined (Mattia Peretti) aillmlocal-llmopen-sourceinference https://blog.mattiaperetti.it/blog/okf-and-the-two-brains-of-an-llm-agent https://blog.mattiaperetti.it/blog/okf-and-the-two-brains-of-an-llm-agent Google quietly published OKF, a plain markdown-plus-frontmatter format for agent knowledge, built on a pattern Karpathy sketched in a gist months earlier. Working through both taught me to stop thinking of an LLM as one brain and start thinking of it as two. Tue, 07 Jul 2026 00:00:00 GMT undefined (Mattia Peretti) aiagentic-aillmmemoryokf https://blog.mattiaperetti.it/blog/the-superpowers-of-a-software-engineer https://blog.mattiaperetti.it/blog/the-superpowers-of-a-software-engineer Coding agents get better every month, and the way to stay in control is to give them clear instructions and context rather than fight the shift. This covers agent instruction file conventions worth adding to a repo (AGENTS.md, SKILL.md, CLAUDE.md, Cursor skills) and two open-source tools, Ponytail and Caveman, that cut token spend by trimming what agents write and compressing what you send. Mon, 13 Jul 2026 00:00:00 GMT undefined (Mattia Peretti) aiagentic-aiclaude-codellm