flutter_gemma library
Classes
- CanceledError
- Download canceled
- CancelToken
- Token for cancelling model downloads
- Condition
- A single predicate over one metadata field. Sealed because the storage codec needs to switch over the concrete subtype — adding a new condition without updating the codec would silently drop it.
- CorruptionDetectionResult
- Result of corruption detection
- DiagnosticReport
- Diagnostic information for troubleshooting
- DocumentWithEmbedding
- Document with its embedding, used to rebuild an in-memory HNSW index.
- DownloadError
- Download error types using sealed classes for exhaustive pattern matching
- DownloadProgress
- Progress information for model downloads
- EmbeddingInstallation
- Result of embedding model installation
- EmbeddingInstallationBuilder
- Fluent builder for embedding model installation
- EmbeddingModel
- Represents an embedding model instance.
- EmbeddingModelSpec
- Specification for embedding models (model.bin + tokenizer.json)
- ErrorHandlingResult
- Result of error handling
- EstimatedDimensions
- Estimated image dimensions
- FieldEquals
-
metadata[key] == valueexact match. - FieldMatchAny
-
metadata[key] in valuesset membership. Equivalent to N FieldEquals wrapped in ashouldbucket, but expressed in one place and serialized more efficiently on the storage side. - FieldRange
-
gte <= metadata[key] <= ltenumeric range. Either bound may be null for one-sided ranges (e.g.gte: 10matches anything >= 10). - Filter
- Payload filter applied alongside vector similarity search.
- FilterField
- A single metadata field a store is told to make filterable.
- FilterSchema
- The set of metadata fields a store should make filterable.
- FlutterGemma
- Modern API facade for Flutter Gemma
- FlutterGemmaDesktop
- Desktop implementation of FlutterGemma plugin
- FlutterGemmaPlugin
- Interface for the FlutterGemma plugin.
- ForbiddenError
- HTTP 403 - Access forbidden
- FunctionCallParser
- Facade for backward compatibility. Delegates to model-specific FunctionCallFormat implementations.
- FunctionCallResponse
- Gemma3Specs
- Gemma 3 SigLIP vision encoder specifications
- GeneralVisionSpecs
- General vision encoder specifications
- ImageErrorHandler
- Comprehensive error handling and debugging utilities for AI image processing to prevent corruption that causes repeating text patterns in model responses.
- ImageProcessor
- Comprehensive image processing utilities to prevent AI image corruption and ensure proper vision encoder compatibility.
- ImageTokenizer
- Handles proper image tokenization for multimodal AI models to prevent "Prompt contained 0 image tokens but received 1 images" errors and corruption that causes repeating text patterns.
- InferenceChat
- InferenceInstallation
- Result of inference model installation
- InferenceInstallationBuilder
- Fluent builder for inference model installation
- InferenceModel
- Represents an LLM model instance.
- InferenceModelSession
- Session managing response generation from the model.
- InferenceModelSpec
- Specification for inference models (main model + optional LoRA)
- Message
- ModelFile
- Represents a single file that belongs to a model
- ModelFileManager
- ModelResponse
- Base interface for model responses from InferenceChat Can be either TextResponse, FunctionCallResponse, or ThinkingResponse
- ModelSpec
- Base specification for any model (inference or embedding)
- MultimodalImageHandler
- Main integration class for handling multimodal image processing in Flutter Gemma to prevent AI image corruption and repeating text pattern issues.
- MultimodalImageResult
- Result of multimodal image processing
- NetworkError
- Network connectivity error
- NotFoundError
- HTTP 404 - Resource not found
- OrphanedFileInfo
- Information about a potentially orphaned file
- ParallelFunctionCallResponse
- Multiple function calls in a single model response.
- ProcessedImage
- Represents a processed image ready for AI model consumption
- RateLimitedError
- HTTP 429 - Rate limit exceeded
- ResponseValidationResult
- Result of response validation
- RetrievalResult
- A single retrieval hit from a vector store query.
- ServerError
- HTTP 5xx - Server error
- SessionMetrics
- Session metrics containing token usage and performance statistics.
- SkillExecutorProvider
-
A pluggable skill executor — the seam between core and the opt-in
flutter_gemma_agentpackage's agentic "skills" runtime. - SkillExecutorRegistry
-
Holds skill executors registered via
FlutterGemma.initialize(skillExecutors:). Same probe-chain selection asEngineRegistry/EmbeddingRegistry: the registered executor with the highest SkillExecutorProvider.priority whose SkillExecutorProvider.canExecute returns true for a skill type wins (first-registered breaks ties). There is no central type map — the opt-influtter_gemma_agentexecutors self-select. - StopTokenFilter
-
Filters stop tokens from model response stream.
For .litertlm on iOS, MediaPipe doesn't handle
<end_of_turn>— this filter detects and terminates the stream at the stop token, with buffering for partial tag matches. - StorageStats
- Storage statistics
- TextResponse
- Text token during streaming
- ThinkingResponse
- Thinking process content from the model
- Tool
- HTTP 401 - Authentication required
- UnknownError
- Unknown error
- ValidationResult
- Result of image validation
- VectorStoreRepository
- Abstract repository for vector store operations
- VectorStoreStats
- Summary statistics for a vector store.
- VisionEncoderValidator
- Validates images for compatibility with AI vision encoders to prevent corruption that causes models to interpret images as repeating text patterns.
- VisionSpecs
- Base class for vision encoder specifications
Enums
- CorruptionAction
- Actions to take when corruption is detected
- ErrorType
- Types of errors that can occur in image processing
- FilterFieldType
- The storage type of a declared filterable metadata field.
- GemmaLogLevel
- Verbosity for flutter_gemma's internal logs.
- MessageType
- ModelFileType
- ModelManagementType
- Base enumeration for different model management types
- ModelReplacePolicy
-
Policy for what happens to a previously-installed model when a new one is
set active. Lives with the spec value types (it's a per-spec install policy);
re-exported from
model_file_manager_interface.dartfor backward compat. - ModelType
- PreferredBackend
- Hardware backend for model inference.
- ResponseAction
- Actions to take for corrupted responses
- TaskType
- Task type for embedding generation, following Google RAG SDK convention.
- ToolChoice
- Controls whether the model should call tools.
- VisionEncoderType
- Vision encoder types with their specifications
- WebStorageMode
- Storage mode for web platform models
Mixins
- RawSdkResponseSession
-
Mixin for sessions that surface the SDK's structured raw JSON response
(LiteRT-LM Gemma 4 path with
tool_calls). Allows InferenceChat to read the structured tool calls without a hard dependency on a concrete session type, and lets non-FFI sessions opt out by simply not implementing this mixin.
Extensions
- DownloadErrorMessage on DownloadError
- Extension to convert DownloadError to user-friendly messages
Constants
- defaultMaxFunctionBufferLength → const int
- Default maximum length for function call buffer before flushing as text. Must accommodate verbose formats (DeepSeek tags, parallel calls).
-
supportedLoraRanks
→ const List<
int>
Exceptions / Errors
- DownloadCancelledException
- Exception thrown when a download is cancelled
- DownloadException
- Custom exception for download failures
- ImageProcessingException
- Exception thrown when image processing fails
- ImageTokenizationException
- Exception thrown when image tokenization fails
- ModelStorageException
- Exception thrown when model storage operations fail
- VectorStoreException
- Exception thrown by VectorStore operations
- VisionEncoderValidationException
- Exception thrown when validation fails