Amazon Translate vs Lokalise

Amazon Translate

IAM policies scope access per action and resource, with a TranslateReadOnly managed policy, and CloudTrail logs each call. AWS may store and use text to improve Amazon AI services unless the organisation sets an opt-out policy.

Lokalise

The REST API has public OpenAPI files, Markdown docs, published rate limits, read-only tokens, scoped OAuth and an audit log endpoint. Access needs a paid plan after a 14-day trial, a person signs up in a browser, and the Master Service Agreement forbids benchmarking. No idempotency keys or deprecation policy were found in the reviewed documentation.

Amazon Translate

1. Chunk real-time text under 10,000 bytes, not characters, so multibyte scripts fit less per call
2. Pass SourceLanguageCode auto to detect the source language, and handle DetectedLanguageLowConfidenceException
3. Read AppliedSettings in the response, since unsupported formality or profanity settings are dropped without an error
4. Retry TooManyRequestsException with backoff, which the AWS SDKs do by default
5. Set the AI services opt-out policy before sending private text

Lokalise

1. Send the token in the X-Api-Token header to https://api.lokalise.com/api2. OAuth access tokens go in Authorization: Bearer and expire after about 60 minutes.
2. Stay under 6 requests a second per token and 10 concurrent requests per project. On 429, back off exponentially and read X-RateLimit-Reset.
3. Use a read-only token for reads. A read/write token can call PUT /projects/{project_id}/empty, which deletes every key and translation.
4. Use pagination=cursor on /keys and /translations, with limit at most 500 on keys, and follow X-Pagination-Next-Cursor.
5. Projects above 10,000 key-language pairs need /files/async-download. Poll /processes/{process_id} for the bundle URL, and do the same after /files/upload.