What's Changed

Bugfixes

• Fix bug passing single-string safety settings. by @MarkDaoust in #417
• Fix operator conversion. by @MarkDaoust in #419
• Fix 'ResourceWarning: unclosed <ssl.SSLSocket>' by @MarkDaoust in #422
• Fix list_models by @rozha in #439
• Add PngImagePlugin to imports. by @mbanani in #438
• Add webp to mimetypes by @MarkDaoust in #440

Docs and Samples

• Caching samples by @shilpakancharla in #399
• Model list and get samples by @shilpakancharla in #404
• System instructions sample by @shilpakancharla in #407
• Count tokens samples by @shilpakancharla in #403
• fix genai.GenerativeModel argument name in example by @philippemiron in #394
• Adding text generation samples by @shilpakancharla in #405
• Add samples for chat by @shilpakancharla in #408
• Rename json_mode.py and create function calling sample by @shilpakancharla in #406
• Add tuned_models samples by @MarkDaoust in #409
• Add no json schema sample by @shilpakancharla in #416
• fix list initialization and redundant parentheses by @PandasPirate in #412
• Adding a note about free tier not available in Europe by @Giom-V in #401
• Update text only prompts by @shilpakancharla in #420
• Add cache_from_name and cache_chat by @shilpakancharla in #413
• Add create_audio and create_text to samples/files.py by @shilpakancharla in #414
• Add safety_settings sample by @shilpakancharla in #415
• Add REST sample for listing and getting models by @shilpakancharla in #421
• Cleanup samples/files.py by @MarkDaoust in #402
• format-samples by @MarkDaoust in #410
• Add model configuration sample by @shilpakancharla in #427
• move SI by @MarkDaoust in #428
• No 's' by @MarkDaoust in #431
• Fix snippet name. by @MarkDaoust in #432
• Fix snippet names + reorder. by @MarkDaoust in #435
• move tuning to tuned_models by @MarkDaoust in #436
• Fix typos by @PandasPirate in #437
• Update tuned_models.py by @MarkDaoust in #441
• Check video status. by @MarkDaoust in #442

New Contributors

• @philippemiron made their first contribution in #394
• @Giom-V made their first contribution in #401
• @rozha made their first contribution in #439
• @mbanani made their first contribution in #438

Full Changelog: v0.7.1...v0.7.2

What's Changed

• Add code execution by @shilpakancharla in #396

Full Changelog: v0.7.0...v0.7.1

What's Changed

For a detailed introduction, check out the context caching guide on the site.

Features

• Explicit Caching by @mayureshagashe2105 in #355
• Explicit Caching patch by @mayureshagashe2105 in #377

Bugfixes and small changes

• Delete .github/ISSUE_TEMPLATE directory by @ryanwilson in #389
• Update gapic lib to use 0.6.5 and fix tests by @mayureshagashe2105 in #390
• Accept partial file names by @mayureshagashe2105 in #386
• Remove duplicate test by @mayureshagashe2105 in #387

Docs

• remove references to pro-vision by @MarkDaoust in #388

New Contributors

• @ryanwilson made their first contribution in #389

Full Changelog: v0.6.0...v0.7.0

What's Changed

Features

• Add genai.protos to discourage direct use of google.ai.generativelanguage by @MarkDaoust in #354
• Add video metadata and error to File by @MarkDaoust in #348
• Restrict Harm category to the subset supported by Gemini by @aertoria in #295
• Improve request_options by @MarkDaoust in #297
• Add GenerateContentResponse.to_dict() by @MarkDaoust in #337
• Quick safety filtering: Allow safety_settings="block_none" by @MarkDaoust in #347

Bugfixes

• Cleanup file data handling. by @MarkDaoust in #321
• Allow count_tokens with no contents. by @MarkDaoust in #342
• improve the no-credentials error message, fail fast for no-credentials in colab. by @MarkDaoust in #352
• Make printing less verbose. by @MarkDaoust in #340
• Add request options to chat. by @MarkDaoust in #341
• Handle image RGBA mode by @MarkDaoust in #374

Docs

• Fix bugs, improve code clarity, and enhance overall reliability across several files. by @Faisal-Alsrheed in #339
• Enhance functions Documentation and Improve Error Messages by @Faisal-Alsrheed in #350
• Fix argument description. by @MarkDaoust in #338
• Update init.py to use the latest model by @logankilpatrick in #362

New Contributors

• @Faisal-Alsrheed made their first contribution in #339
• @logankilpatrick made their first contribution in #362

Full Changelog: v0.5.4...v0.6.0

What's Changed

• Update version of generativelanguage to fix timeout errors. by @MarkDaoust in #345

What's Changed

Features

• Extend permissions service support to Tuned Models by @mayureshagashe2105 in #219
• Support improved CountTokens and usage_metadata output by @MarkDaoust in #315
• Add JSON schema support
  • Add response_schema parameter by @shilpakancharla in #302
  • Improve schema support by @MarkDaoust in #309
• Upgrade file handling
  • Add file state by @MarkDaoust in #316
  • Add support for resumable uploads for File API by @TYMichaelChen in #275
  • Return full file details at the end of create. by @MarkDaoust in #314
    • Update client.py by @MarkDaoust in #325

Bugfixes

• update_time defined twice hence removing once. by @PandasPirate in #301

• Add main async code match test by @mayureshagashe2105 in #299

• Copy test cases over for new responder.py by @shilpakancharla in #303

• Add f to f-string by @markmcd in #322

Docs

• Debug docs build. by @MarkDaoust in #298
• Update issue templates to point to the Discourse forum by @ymodak in #304
• typo by @MarkDaoust in #313

New Contributors

• @PandasPirate made their first contribution in #301

Full Changelog: v0.5.2...v0.5.3

Bug Fix

• Fix file uploads by @MarkDaoust in #291

New Features

• Support for new TaskTypes and output_dimensionality param by @mayureshagashe2105 in #285

Full Changelog: v0.5.1...v0.5.2

New features

• Add JSON mode by @shilpakancharla in #274

Docs and small fixes

• Fix: incorrect type checking by @Yeuoly in #283
• Fix an internal test failure by @markmcd in #279
• Allow incomplete GenerationConfigDict by @markmcd in #286
• Use generativelanguage v0.6.2 for json mode + update package version. by @MarkDaoust in #287

New Contributors

• @Yeuoly made their first contribution in #283

Full Changelog: v0.5.0...v0.5.1

What's Changed

New features

• Files API

  You can now upload files to the api. Then instead of resending an image with each request, just pass a file reference in your prompt:

  f = genai.upload_file(path="image.png")
  m = genai.GenerativeModel(model_name=...)
  response = m.generate_content(["Please describe this file:", f])
  

  • Add initial prototype supporting the files API by @MarkDaoust in #249
  • Files API: improve error messages, add delete_file, add .uri property by @TYMichaelChen in #260

• Semantic retriever

  Using answer.generate_answer, you can now either pass "inline_passages" to include the reference material in the request:

  from google.generativeai import answer
  answer.generate_answer(
      content=question,
      inline_passages=splitter.split(document)
  )
  

  Or pass a reference to a retriever Document or Corpus:

  from google.generativeai import answer
  from google.generativeai import retriever
  my_corpus = retriever.get_corpus('my_corpus')
  genai.generate_answer(
      content=question,
      semantic_retreiver=my_corpus
  )
  

  • Finish Semantic Retriever. by @shilpakancharla in #230

• System instructions

  When creating a model you can pass a string (or Content) as system_instructions: genai.GenerativeModel(system_instructions="Be good!")

  • Add basic SI and tool_config support by @MarkDaoust in #257
  • Add system_instruction to model repr by @markmcd in #272

• Function Calling - Tool Config

  The list of tools available for function calling during a chat session is typically constant. The new tool_config argument lets you switch the function calling mode between None (No function calls), Auto (The model chooses to return text or a FunctionCall), and Any (the model must call a function).

  model = genai.GenerativeModel(tools=[my_functions])
  chat = model.start_chat()
  chat.send_message(tool_config={"function_calling_config": "ANY"})
  

  In Any mode the model will return a function call. You may also pass allowed_function_names to restrict the calls to a subset of the available functions: `chat.send_message(tool_config={"function_calling_config": "ANY", "allowed_function_names": [...]})

  In Any mode the model can use constrained decoding to more strictly follow the argument specifications in the function declarations.

  • Add basic SI and tool_config support by @MarkDaoust in #257
  • Function calling mode patch by @mayureshagashe2105 in #271

Docs, bug fixes, and minor updates.

• Update README.md by @aertoria in #256
• Fix incorrect reference to self.candidates[0].parts, self.candidates[0].parts[0].text by @Andy963 in #239
• Fix TypedDict usage for compatibility with PyDantic by @MarkDaoust in #259
• Simplify the Python SDK README by @random-forests in #267
• Sdk update to use glm 0.6.1 by @mayureshagashe2105 in #270

New Contributors

• @aertoria made their first contribution in #256
• @TYMichaelChen made their first contribution in #260
• @random-forests made their first contribution in #267

Full Changelog: v0.4.1...v0.5.0

Bugfix

• The send_message_async method should call the asynchronous method generate_content_async by @xnny in #229

Docs

• Fix wrong arg name in README by @lll-lll-lll-lll in #228

New Contributors

• @lll-lll-lll-lll made their first contribution in #228
• @xnny made their first contribution in #229

Full Changelog: v0.4.0...v0.4.1