v3.7.0

Added

• Added support for listing translation memories via DeepL.translation_memories.list.
• Added translation_memory and translation_memory_threshold parameters to translate()
  to use a translation memory during text translation. The translation_memory parameter
  accepts a string ID or a TranslationMemory object.
• Added support for style rules CRUD operations via DeepL.style_rules:
  create(), find(), update_name(), update_configured_rules(), and destroy().
• Added support for style rule custom instruction CRUD operations via DeepL.style_rules:
  create_custom_instruction(), find_custom_instruction(),
  update_custom_instruction(), and destroy_custom_instruction().
  Please refer to the README for usage instructions.

Changed

• Updated unit test to expect the model_type_used to now be quality_optimized
• Set explicit image for gitlab release CI job to avoid missing $HOME issues

v3.6.1

Fixed

• Fixed broken 3.6.0 release

From 3.6.0:

Added

• Added tag_handling_version parameter to translate() to specify which version of the tag handling algorithm to use. Options are v1 and v2.

Fixed

• extra_body_parameters will no longer cast values to string, as this library sends JSON-encoded requests
  (allows e.g. sending booleans)

v3.5.1

Fixed

• Fixed broken 3.5.0 release

From 3.5.0:

Added

• Added custom_instructions parameter to translate() to customize translation
  behavior with up to 10 instructions (max 300 characters each). Only supported for
  target languages: de, en, es, fr, it, ja, ko, zh and their variants.
  Note: using the custom_instructions parameter will use the quality_optimized
  model type as the default. Requests combining custom_instructions and the
  latency_optimized model type will be rejected.

v3.4.1

Fixed

• Fixed broken 3.4.0 release

From 3.4.0:

Added

• Added support for the GET /v3/style_rules endpoint in the client library, the
  implementation can be found in the StyleRuleApi class. Please refer to the
  README for usage instructions
• Added style_id option to translate() which allows text translation with
  style rules.

Fixed

• Send correct library version in user-agent. Expose library version in the module as a variable.

v3.5.0

[3.5.0] - 2025-12-03

Added

• Added custom_instructions parameter to translate() to customize translation behavior with up to 10 instructions (max 300 characters each). Only supported for target languages: de, en, es, fr, it, ja, ko, zh and their variants. Note: using the custom_instructions parameter will use the quality_optimized model type as the default. Requests combining custom_instructions and the latency_optimized model type will be rejected.

v3.3.0

Added

• Added extra_body_parameters option to text and document translation methods to pass arbitrary parameters in the request body. This can be used to access beta features or override built-in parameters (such as target_lang, source_lang, etc.).
• Added rephrase functionality to the top-level DeepL module for convenience.
  • Thanks to 341bms in #14 for the report!

v3.2.0

Added

• Added rephrase functionality to get a new version of submitted text with various possible styles or tones applied
• Added DeepL::Constants namespace and associated constant values for options possibilities

v3.1.0

Added

• Added model_type option to translate() to use models with higher
  translation quality (available for some language pairs), or better latency.
  Options are 'quality_optimized', 'latency_optimized', and 'prefer_quality_optimized'

• Added the model_type_used field to the translate() response, that
  indicates the translation model used when the model_type option is
  specified.

v3.0.2

Added

• Added doc example and tests for context parameter

Fixed

• Fix metadata displayed on RubyGems.org for this library.
• Fixed library version sent in the User-Agent string.

v3.0.1

Fixed

• document.translate_document required a filename, this is now optional. The example in the README now works.