|
1 | 1 | # Unreleased
|
2 |
| -* [fixed] **Breaking Change**: Fixed missing builder methods and return types in builders. |
3 |
| -* [changed] **Breaking Change**: `LiveModelFutures.connect` now returns `ListenableFuture<LiveSessionFutures>` instead of `ListenableFuture<LiveSession>`. |
4 |
| - * **Action Required:** Remove any transformations from LiveSession object to LiveSessionFutures object. |
5 |
| - * **Action Required:** Change type of variable handling `LiveModelFutures.connect` to `ListenableFuture<LiveSessionsFutures>` |
6 |
| -* [changed] **Breaking Change**: Removed `UNSPECIFIED` value for enum class `ResponseModality` |
7 |
| - * **Action Required:** Remove all references to `ResponseModality.UNSPECIFIED` |
8 |
| -* [changed] **Breaking Change**: Renamed `LiveGenerationConfig.setResponseModalities` to `LiveGenerationConfig.setResponseModality` |
9 |
| - * **Action Required:** Replace all references of `LiveGenerationConfig.setResponseModalities` with `LiveGenerationConfig.setResponseModality` |
10 |
| -* [feature] Added support for `HarmBlockThreshold.OFF`. See the |
11 |
| - [model documentation](https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/configure-safety-filters#how_to_configure_content_filters){: .external} |
12 |
| - for more information. |
13 |
| -* [fixed] Improved thread usage when using a `LiveGenerativeModel`. (#6870) |
14 |
| -* [fixed] Fixed an issue with `LiveContentResponse` audio data not being present when the model was |
15 |
| - interrupted or the turn completed. (#6870) |
16 |
| -* [fixed] Fixed an issue with `LiveSession` not converting exceptions to `FirebaseVertexAIException`. (#6870) |
17 |
| -* * [changed] **Breaking Change**: Removed the `LiveContentResponse.Status` class, and instead have nested the status |
18 |
| - fields as properties of `LiveContentResponse`. (#6906) |
19 |
| -* [changed] **Breaking Change**: Removed the `LiveContentResponse` class, and instead have provided subclasses |
20 |
| - of `LiveServerMessage` that match the responses from the model. (#6910) |
21 |
| -* [feature] Added support for the `id` field on `FunctionResponsePart` and `FunctionCallPart`. (#6910) |
22 |
| -* [feature] Add support for specifying response modalities in `GenerationConfig`. (#6921) |
23 |
| -* [feature] Added a helper field for getting all the `InlineDataPart` from a `GenerateContentResponse`. (#6922) |
| 2 | +* [feature] Initial release of the Firebase AI SDK (`firebase-ai`). This SDK *replaces* the previous |
| 3 | + Vertex AI in Firebase SDK (`firebase-vertexai`) to accommodate the evolving set of supported |
| 4 | + features and services. |
| 5 | + * The new Firebase AI SDK provides **Preview** support for the Gemini Developer API, including its |
| 6 | + free tier offering. |
| 7 | + * Using the Firebase AI SDK with the Vertex AI Gemini API is still generally available (GA). |
24 | 8 |
|
| 9 | + If you're using the old `firebase-vertexai`, we recommend |
| 10 | + [migrating to `firebase-ai`](/docs/ai-logic/migrate-to-latest-sdk) |
| 11 | + because all new development and features will be in this new SDK. |
| 12 | + |
| 13 | +* [feature] **Preview:** Added support for specifying response modalities in `GenerationConfig` |
| 14 | + (only available in the new `firebase-ai` package). This includes support for image generation using |
| 15 | + [specific Gemini models](/docs/vertex-ai/models). |
| 16 | + |
| 17 | + Note: This feature is in Public Preview, which means that it is not subject to any SLA or |
| 18 | + deprecation policy and could change in backwards-incompatible ways. |
0 commit comments