Model Effort Levels

Anthropic's configurable reasoning budget hierarchy for Claude models. Controls how many reasoning tokens the model allocates per turn.

Hierarchy

none → low → medium → high → max → X high (Opus 4.7 exclusive)

Key points

Effort levels control thinking depth, not model weights ^[src-010]
Opus 4.6 was silently set to "medium" default on 2026-02-09 via adaptive thinking parameter changes — not a model weight change. This caused the widespread reports of Opus 4.6 degradation ^[src-010]
"X high" is exclusive to Opus 4.7 — Anthropic's stated maximum reasoning allocation ^[src-010]
Higher effort = more thinking tokens = more expensive. The "ultra think" keyword in Claude Code allocates maximum thinking budget (~32K tokens) before responding ^[src-011]
Community guidance: use "X high" / "ultra think" for architecture decisions and hard debugging; use lower effort for routine tasks to preserve quota ^[src-011]

Related entities

Claude Opus 4 7 — introduces X high level
Anthropic — defines and controls the effort hierarchy

Related concepts

Adaptive Thinking — mechanism that selects effort level per turn
Token Economics — effort level is the primary lever on reasoning token spend

Source references

^[src-010] Nate Herk — Cloud agents & model releases cluster (2026-04-14 to 2026-04-17)
^[src-011] Nate Herk — Claude Code power features cluster (2026-04-20 to 2026-04-27)

Robin Cartier perspective

This page is part of Robin Cartier's working AI knowledge graph: a practical research layer for production AI, recommendation systems, experimentation, GEO, and agentic web readiness.

The useful next step is to connect this concept back to applied product leadership and operating models.

Recommended next

Keep reading from this thread

From 500 indexed pages and articles.