My current stack, favorite models, and the tools I actually use every day. Updated as things change, which is constantly.
Last updated: August 2026
01My favorite model. I use Fable 5 for prolonged, complex repository work, deep reviews, ambitious implementations, and tasks where judgment matters as much as raw coding ability.
Best for long-horizon implementation and review
Excellent product and UI judgment
Catches subtle issues other models miss
Strong at maintaining intent across complex work
Extremely expensive through the API
Safeguards and model routing can make availability unpredictable
02My second most-used model and the dependable choice for serious technical work. Deliberate, persistent, and strong at understanding large codebases before making changes.
Excellent for complex, multi-step engineering work
Strong codebase comprehension
Persistent across long-running tasks
More efficient and capable than GPT-5.5
Can be slower and more methodical than necessary
Product and UI taste still trails Fable 5
03My third most-used model. Fast, capable, and particularly good at coding, tool use, and agentic work. I especially like using it through the Grok Build CLI.
Excellent coding and tool calling
Fast and direct
Strong performance for its price
Works especially well inside Grok Build
Not my first choice for the hardest repository work
Newer ecosystem than OpenAI or Anthropic