v0.9.0

0.9.0

• Simplify function calling and remove provided function execution.
• Add code execution tool.

implementation("com.google.ai.client.generativeai:generativeai:0.9.0")

https://maven.google.com/web/index.html?q=generative#com.google.ai.client.generativeai:generativeai:0.9.0

v0.8.0

0.8.0

• Make CountTokens contain the full request
• Fix issue deserializing some function call responses (Issue #160 )

v0.7.0

0.7.0

• Added responseSchema to GenerationConfig
• Added responseMimeType sample usage to refdoc of GenerationConfig

v0.6.0

0.6.0

• Add GenerateContentResponse#functionCalls and mark GenerateContentResponse#functionCall deprecated
• Add support for responseMimeType
• Fix proguard configuration for release builds (#139)

v0.5.0

0.5.0

• Remove usages of v1beta opt-in flag as the default api version is "v1beta"
• Expose UsageMetadata in GenerateContentResponse
• Fix parsing of UsageMetadata to address #134

v0.4.0

0.4.0

• license field is now optional (nullable) in Citations
• The default api version is now v1beta
• Files uploaded using the File service can now be referenced in the prompt using the new FileDataPart
• Implement error catching for unsupported Part types.

v0.3.0

0.3.0

• Add function calling
• Make totalTokens in CountTokensResponse nullable
• Keep argument positions
• Implement error catching for unsupported Part types.

v0.2.2

0.2.2

• Fix issue causing Chat messages to fail due to missing role.

v0.2.1

0.2.1

• Make "user" the default role in all requests
• Add InvalidAPIKeyException
• Add UnsupportedUserLocationException

v0.2.0

0.2.0

• Add RequestOptions; configuration points for backend implementation details such as api version and timeout.
• Support a general model naming schema
• GenerativeModelFutures now correctly passes the prompt through