effort-calibrator

# Effort Calibrator On Fable 5, effort is the primary dial trading intelligence against latency and cost. Defaults that were right for earlier models are usually wrong here: Fable 5 at lower effort frequently beats earlier models at `xhigh`, so over-provisioning effort wastes money and time without adding quality. ## Starting points by workload | Workload | Start at | |---|---| | Routine transforms, classification, short edits, chat, subagents | `medium` (try `low` if latency matters) | | Most analysis and writing | `high` (the general default) | | Coding and agentic/tool-heavy work | `high` (the API and Claude Code default on Fable 5) — even for workloads that ran at `xhigh` on earlier Opus models; escalate to `xhigh` only for the most capability-sensitive tasks | | Hardest capability-sensitive work: large migrations, multi-day autonomous runs, novel research | `xhigh` | | Frontier problems only, where evals show headroom above `xhigh` and token spend is unconstrained | `max` | The signal for `max` is evals showing headroom above `xhigh` on your actual task: on most workloads it adds significant cost for small gains and can tip into overthinking. ## Adjustment signals Lower effort when: - Tasks complete correctly but take longer than the work warrants - The session is interactive and waiting hurts more than marginal quality helps - Output shows over-deliberation: long context-gathering before trivial actions Raise effort when: - First-shot correctness matters more tha...