Effort Levels Simplified to Low / Medium / High with New Visual Indicators
Claude CodeClaude Code 2.1.72 overhauls the effort level system by consolidating it from four tiers down to three — low, medium, and high — and retiring the "max" tier. Each level now has a distinct symbol (○ ◐ ●) for at-a-glance visibility, and the persistent status icon is replaced with a brief notification on change. The /effort auto command resets the setting back to the model's default, giving developers a clean escape hatch.
Sources & Mentions
4 external resources covering this update
A Cleaner Effort Dial
Claude Code's effort system controls how deeply Claude reasons through a task — higher effort means more extended thinking, more thorough analysis, and better results at the cost of latency and tokens. The system has accumulated tiers over time, and v2.1.72 simplifies it into three clearly defined levels:
| Symbol | Level | Best for |
|---|---|---|
| ○ | Low | Fast, lightweight tasks — lookups, reformatting, simple completions |
| ◐ | Medium | Most everyday coding work — the default sweet spot |
| ● | High | Complex reasoning, architectural decisions, hard bugs |
What Changed
The previous "max" tier is retired. Developers who relied on maximum effort for demanding tasks should use high going forward, which now serves as the ceiling. The UX also changes: instead of a persistent icon occupying the status line, effort changes now surface as a brief notification on change — less visual clutter during active sessions.
The /effort auto command is new and resets the effort setting to the model's default. Previously there was no clean way to undo a manual /effort command without knowing what the default was; now a single command returns to automatic selection.
Why It Matters
Effort levels have become increasingly central to how developers manage Claude's cost and speed tradeoffs. Removing the ambiguity between "high" and "max" reduces decision fatigue and makes the dial easier to reason about. The new symbols are readable in a terminal without relying on text labels, which is practical when effort state needs to be monitored at a glance.
The simplification also aligns with Anthropic's direction set in v2.1.68, where Opus 4.6 was switched to medium effort by default for Max and Team subscribers — the three-tier model makes that default-setting story cleaner.
Additionally, the --effort CLI flag is fixed in this release to stop being silently reset by unrelated settings writes on startup, ensuring the flag persists as intended across a session.