What's Changed

• Add default RequestOptions.timeout of 300 seconds by @andrewheard in #170
• Align text and add close keyboard function in sample app by @1998code in #172
• Update models in samples and README to gemini-1.5-flash-latest by @andrewheard in #173
• Add responseSchema to GenerationConfig by @andrewheard in #176
• Send GenerateContentRequest in CountTokensRequest by @andrewheard in #175
• Log server error responses without -GoogleGenerativeAIDebugLogEnabled by @andrewheard in #177

New Contributors

• @1998code made their first contribution in #172

Full Changelog: 0.5.3...0.5.4

What's Changed

• Add usageMetadata to GenerateContentResponse by @andrewheard in #159
• Re-order the GenerativeModel convenience constructors in source by @andrewheard in #160
• Add responseMIMEType to GenerationConfig by @andrewheard in #161
• Add ModelContent tests for fileData parts by @andrewheard in #163
• Update doc comments by @morganchen12 in #166
• Make text computed property handle mixed-parts responses by @andrewheard in #165

Full Changelog: 0.5.2...0.5.3

What's Changed

• Make system instruction accept String values instead of ThrowingPartsRepresentable by @andrewheard in #152, #155
  • Only text values are supported for system instructions.
• Add default nil value for requiredParameters in FunctionDeclaration by @andrewheard in #153

Full Changelog: 0.5.1...0.5.2

What's Changed

• Allow for a simpler way to set system instructions by @peterfriese in #149

Full Changelog: 0.5.0...0.5.1

What's Changed

• Update Citation decoding to handle optional values by @andrewheard in #135
• Add URI-based file data support by @andrewheard in #134
• Update default API version from v1 to v1beta by @andrewheard in #137
• Switch to gemini-1.5-pro-latest in docs and tests by @andrewheard in #138
• Make Citation.license optional by @andrewheard in #141
• Remove public Codable (Decodable and/or Encodable) conformance by @andrewheard in #142

Breaking Changes

• Citation.license is now optional (String?)
• ModelContent.Part adds a new fileData enum case

Full Changelog: 0.4.11...0.5.0

What's Changed

• Add function calling sample by @andrewheard in #127
• Add system instruction support by @andrewheard in #129

Full Changelog: 0.4.10...0.4.11

What's Changed

• Add functionCalls accessor to GenerateContentResponse by @andrewheard in #123
• Forced function calling by @paulb777 in #124
• Increment SDK version to 0.4.10 by @paulb777 in #126

Full Changelog: 0.4.9...0.4.10

What's Changed

• Allow PartsRepresentable to throw errors by @morganchen12 in #88
• Update ModelContent.Part enum decoding strategy by @andrewheard in #115
• Capture and share the backend's message for an invalid API key by @paulb777 in #120
• Add function calling support by @andrewheard in #116

Full Changelog: 0.4.8...0.4.9

What's Changed

• Add RequestOptions to the GenerativeModel constructor, which allows developers to specify configuration options for requests sent to the backend (#100)
  • apiVersion: Optionally target a specific version of the API (for example, "v1beta"); defaults to "v1" if not specified (#101)
  • timeout: Optionally set a request’s timeout interval in seconds; if not specified, uses the default value for a URLRequest (#99)
• The PaLM API is deprecated for use with Google AI services and tools (but not for Vertex AI). Learn more about this deprecation, its timeline, and how to migrate to use Gemini in the PaLM API deprecation guide (#110)
• Update model name strings in README, docs and samples; see Gemini models for a list of supported model names (#109)
• Add support for any model name prefix, in addition to models/ (#98)
• Decrease minimum required Swift toolchain version to 5.7.1 (#102, #103)

Full Changelog: 0.4.7...0.4.8

What's Changed

• Resize images to 768x768px in GenerativeAIMultimodalSample by @andrewheard in #75
• Add invalidAPIKey case to GenerateContentError by @andrewheard in #91, #92, #93
• Add unsupportedUserLocation case to GenerateContentError by @andrewheard in #85
• Add new error case handling in the Chat Sample by @andrewheard, @peterfriese in #94, #95

Full Changelog: 0.4.6...0.4.7