Claude Code runs on Anthropic's Claude models, so the real question is which one. Claude Opus 4.8 is the pick for hard, high-stakes tasks; Claude Sonnet 4.6 is the everyday default that most users keep selected; and Claude Haiku 4.5 is the cheap, fast option for subagents and high-volume steps.
Asking which Claude model is best overall, not just inside Claude Code? Best Claude Model in 2026 →
The quality ceiling for Claude Code — reach for it on hard refactors, gnarly bugs, and tasks where one good run beats three cheap ones.
The everyday default: most of Opus's coding ability, roughly half the price, and faster — what most Claude Code sessions should be on.
For subagents and the many cheap calls a session makes — fast and inexpensive, with the same Claude reliability.
The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.
Every model on this page is included with Standard Compute
