---
title: "AI content: translate the whole site into one more language (saved as a version; multilingual sites are an add-on). Metered in Build Credits (DeepSeek, a few hundredths of a credit for a short text); the answer includes usage."
method: POST
path: "/builder/sessions/{sessionId}/content/languages"
tags: ["Builder"]
---

# AI content: translate the whole site into one more language (saved as a version; multilingual sites are an add-on). Metered in Build Credits (DeepSeek, a few hundredths of a credit for a short text); the answer includes usage.

`POST /builder/sessions/{sessionId}/content/languages`

## Request

- Base URL: `https://hostingsurge.com/api/v1`
- URL: `https://hostingsurge.com/api/v1/builder/sessions/{sessionId}/content/languages`
- Auth: one of:
  - HTTP bearer
  - API key in cookie `__Host-hs_session`

## Path parameters

- `sessionId` string, uuid, required

## Request body

- object
  - `lang` string, required
  - `formality` 'informal' | 'formal'
  - `glossary` string[]: Words never translated (brand and product names); saved for later updates
  - `translateAddresses` boolean: Give the pages addresses in the new language (default: yes for Latin-script languages)

## Response `201`

The version

- object
  - `turn` object, nullable, required: The version (null when nothing changed)
    - `id` string, uuid, required
    - `seq` integer, required
    - `prompt` string, required
    - `status` 'queued' | 'running' | 'done' | 'failed' | 'canceled', required
    - `summary` string, nullable, required: The assistant's chat reply for this turn
    - `model` string, required: The model that produced the final attempt (the stronger model when the turn escalated)
    - `escalated` boolean, required: Whether the turn used the stronger (escalation) model
    - `escalation` 'auto' | 'user' | 'null', nullable, required
    - `attempt` integer, required: Model calls used, including auto-fix retries
    - `usage` object, required
      - `inputTokens` integer, required
      - `outputTokens` integer, required
      - `cachedInputTokens` integer, required: Input tokens served from the provider's prompt cache (included in inputTokens)
      - `costEstimateMicros` integer, required: Model cost in micro-USD (1,000,000 = $1): the tokens of every model call × the configured price of the model it used
      - `millicredits` integer, required: Build Credits × 1000 this turn costs (rounded up per turn)
      - `credits` number, required: Build Credits this turn costs (millicredits / 1000); metered, charged only once billing is enabled
      - `byModel` object[], required: Usage per model (the default and, after an escalation, the stronger model)
        - `model` string, required
        - `attempts` integer, required: Model calls made with this model
        - `inputTokens` integer, required
        - `outputTokens` integer, required
        - `cachedInputTokens` integer, required
        - `costMicros` integer, required: Model cost in micro-USD (1,000,000 = $1)
    - `filesChanged` integer, required
    - `changedFiles` object[]: Builder v2: the files this turn changed, so a client can rebuild the chat after a reload without fetching diffs
      - `path` string, required
      - `change` 'added' | 'modified' | 'deleted', required
    - `commitSha` string, nullable, required
    - `kind` 'chat' | 'manual' | 'restore' | 'code' | 'github' | 'merge' | 'start': What made the version: chat (default), manual (edits made in the preview), restore, code (code mode), github (from the linked GitHub branch), merge (a draft kept) or start
    - `editCount` integer, nullable: manual: the number of preview edits this version collects
    - `bookmarked` boolean: The customer bookmarked this version as a stable one
    - `error` object, nullable, required
      - `code` string, required
      - `message` string, required
    - `qa` object, nullable: The automatic browser test of the result of this turn (null: none ran; migration 197)
      - `status` 'passed' | 'fixed' | 'issues' | 'skipped', required: passed: nothing to fix; fixed: every problem found was fixed by the builder; issues: problems remain; skipped: no test ran
      - `reason` 'unavailable' | 'not_running' | 'no_browser' | 'failed' | 'time' | 'canceled' | 'null', nullable, required: skipped: why (the preview has no browser test on this server, the preview did not run, the test failed or ran out of time)
      - `pages` integer, required: Pages tested
      - `viewports` integer[], required: Widths tested, in CSS pixels (1280 = desktop, 390 = phone)
      - `clicks` integer, required: Buttons and links clicked
      - `forms` integer, required: Forms filled in and sent in a dry run (nothing leaves the preview)
      - `passes` integer, required: Browser tests run in this turn
      - `found` integer, required: Problems the first test found
      - `fixed` integer, required: Of those, fixed by the builder before the turn ended
      - `remaining` object[], required: Problems the last test still found (most important first)
        - `id` string, required: Stable id of the problem (the same problem keeps its id between tests)
        - `kind` 'exception' | 'console_error' | 'request_failed' | 'page_not_found' | 'broken_link' | 'broken_anchor' | 'control_error' | 'form_error' | 'a11y' | 'contrast' | 'visual' | 'page_error', required
        - `severity` 'error' | 'warning', required: error: the builder fixes it by itself; warning: shown, not fixed automatically
        - `page` string, nullable, required: The page (path) where it happened
        - `viewport` 'desktop' | 'mobile' | 'both' | 'null', nullable, required
        - `target` string, nullable, required: What: the label of a button, a link, a file or a rule
        - `detail` string, required: The exact message, in English (page text: untrusted)
        - `count` integer, required
      - `remainingCount` integer, required
      - `stopped` 'attempts' | 'no_progress' | 'time' | 'budget' | 'canceled' | 'failed' | 'null', nullable, required: Why the builder stopped fixing while problems remained: its attempts for this message were used up, a fix changed nothing, the time or the spending limit for automatic fixes was reached, the turn was stopped, or the fix or its test could not be finished (failed)
      - `visual` 'off' | 'done' | 'failed', required: The vision check of the screenshots (only where it is switched on)
      - `durationMs` integer, required
    - `createdAt` string, required: ISO-8601 timestamp
    - `startedAt` string, nullable, required: ISO-8601 timestamp
    - `finishedAt` string, nullable, required: ISO-8601 timestamp
  - `previousTurnId` string, nullable, required: Restore this version to undo the change
  - `changedFiles` string[], required
  - `untranslated` string[], required: Texts that keep the default language's words for now (shown as "needs update")
  - `usage` object, nullable, required
    - `model` string, required
    - `calls` integer, required
    - `inputTokens` integer, required
    - `outputTokens` integer, required
    - `cachedInputTokens` integer, required
    - `costMicros` integer, required: Model cost, micro-USD
    - `millicredits` integer, required
    - `credits` number, required
    - `charge` 'charged' | 'duplicate' | 'skipped' | 'failed' | 'none', required: What billing did with it (skipped: billing is off or the team is unlimited)

## Other responses

- `400`: Invalid request (code invalid_request; details.reason: invalid, unmappable, not_found, secret_detected)
- `401`: Missing or invalid session/token (code unauthorized)
- `402`: No Build Credits left, the team's daily credits are used (code limit_reached), or multilingual sites need a plan or credits (code upgrade_required, details.feature site.multilingual)
- `403`: Not permitted, or the AI builder is in private beta (code forbidden)
- `404`: Not found or not visible to the caller (code not_found)
- `409`: Refused for now (code conflict; details.reason turn_running, archived, language_exists, no_content, stale)
- `429`: Too many AI actions (code rate_limited; details.reason content_rate or capacity); see Retry-After
- `501`: The AI is not available on this server right now (code not_configured)
- `503`: The AI could not finish (code provider_unavailable); nothing was changed

---

[API](https://skmtc.dev/hostingsurge/apis/hostingsurge-platform-api.md) · [All operations](https://skmtc.dev/hostingsurge/apis/hostingsurge-platform-api/llms.txt) · [OpenAPI document](https://skmtc.dev/hostingsurge/apis/hostingsurge-platform-api/revisions/4d08337cdff1?raw)