AI models, product engineering, and the tools I ship with.
11
Recent writing tagged Models.
I'm scrapping my custom RAG pipelines and wrapper layers to go all-in on Claude Fable 5’s native agentic architecture and 5M token context.
Daniel Ensminger explains why Google's new thinking_level API and massive 64K token output limit in Gemini 3.1 Pro simplifies AI reasoning.
I’m switching my default model to Sonnet 4.6 after seeing its flagship-level engineering performance at a fraction of the cost.
OpenAI's GPT-5.3-Codex marks a paradigm shift from simple autocomplete to a state-manipulating agent that actually understands the terminal.
Archive
1 post
Claude Opus 4.6 just dropped with a 1-million token context window and adaptive thinking, forcing a total rethink of how I build production AI agents.
By subscribing you consent to email updates. You can unsubscribe anytime. See our Privacy Policy.
Follow me on X
