Estimate a CLI v0 run
Price the pending legacy delta or a deliberate refresh without translating its content.
lingo run --estimate calculates selected translation work and sends its character counts to the Lingo.dev pricing backend. In 0.138.10, it does not translate, write target output, or bill a translation run. It still makes a backend request.
Match the estimate to the intended run #
From the directory owning i18n.json, estimate ordinary pending work:
lingo run --estimateThe delta uses existing targets and i18n.lock. Missing keys and changed source can produce different quantities for each locale. Zero pending work does not establish that present translations are correct.
If you intend to refresh Spanish regardless of source changes, include force in both the estimate and later run:
lingo run --estimate --force --target-locale esA full-project --estimate --force estimates broader work. Do not compare an incremental estimate with a forced run as though they describe the same request.
Interpret the result #
Output reports source characters, approximate output tokens and cost per target locale, plus a total. Output tokens use a character-to-token heuristic, so actual model usage and translation cost can differ. This is an estimate, not a quote or measured saving.
V0 estimation requires the Lingo.dev backend. It is not a pricing interface for direct OpenAI, Anthropic, or Ollama use, and --estimate cannot combine with --watch or --frozen.
Keep the estimate with its source revision, selected locales, localization engine, and force setting. Repeat it if those inputs or pricing change. Preserve human-edited targets before retranslation; an estimate does not protect them from replacement.
The current CLI estimate has a different contract: it uploads source bytes to plan work. Do not infer its data flow from this v0 count-based request.