1 · Current stack rank (September 2025)
| Rank | Model | Why we reach for it |
|---|---|---|
| 1 | Claude Opus 4.1 | Highest reliability on complex planning, architecture decisions, and long edits. Slight lead in reasoning and code quality, but also the highest cost. |
| 2 | GPT‑5 Codex | Nearly Opus-level output with noticeably lower latency and ~5× lower cost. Great daily driver for implementation work. |
| 3 | GPT‑5 Claude Sonnet 4 | Solid generalists. We see similar behavior between them; pick based on preference, latency, or cost. |
We ship model updates regularly. When a new release overtakes the list above,
we update this page and the CLI defaults.
2 · Match the model to the job
| Scenario | Recommended model |
|---|---|
| Deep planning, architecture reviews, ambiguous product specs | Start with Opus 4.1. Switch down if you only need execution after the plan is locked. |
| Full-feature development, large refactors | GPT‑5 Codex balances quality and speed; Opus is a good fallback if Codex struggles. |
| Repeatable edits, summarization, boilerplate generation | GPT‑5 or Sonnet 4 keep costs low while staying accurate. |
| CI/CD or automation loops | Favor GPT‑5 Codex or Sonnet 4 for predictable throughput. Promote the critical planning steps to Opus 4.1 when correctness matters most. |
/model or by toggling in the settings panel (Shift+Tab → Settings).
3 · Switching models mid-session
- Use
/model(or Shift+Tab → Settings → Model) to swap without losing your chat history. - If you change providers (e.g. Anthropc to OpenAI), the CLI converts the session transcript between Anthropic and OpenAI formats. The translation is lossy—provider-specific metadata is dropped—but we have not seen accuracy regressions in practice.
- For the best context continuity, switch models at natural milestones: after a commit, once a PR lands, or when you abandon a failed approach and reset the plan.
- If you flip back and forth rapidly, expect the assistant to spend a turn re-grounding itself; consider summarizing recent progress when you switch.
4 · Reasoning effort settings
- Anthropic models (Opus/Sonnet) show modest gains between Low and High.
- GPT models respond much more to higher reasoning effort—bumping GPT‑5 or GPT‑5 Codex to High can materially improve planning and debugging.
- Reasoning effort increases latency and cost, so start Low for simple work and escalate when you need more depth.
Change reasoning effort from
/model → Reasoning effort, or via the
settings menu.5 · Bring Your Own Keys (BYOK)
Factory ships with managed Anthropic and OpenAI access. If you prefer to run against your own accounts, BYOK is opt-in—see Bring Your Own Keys for setup steps, supported providers, and billing notes.6 · Keep notes on what works
- Track high-impact workflows (e.g., spec generation vs. quick edits) and which combinations of model + reasoning effort feel best.
- Ping the community or your Factory contact when you notice a model regression so we can benchmark and update this guidance quickly.