> ## Documentation Index > Fetch the complete documentation index at: https://docs.mixpeek.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # Partially Update Cluster > This endpoint partially updates a cluster (PATCH operation). Only provided fields will be updated. At minimum, metadata can always be updated. Immutable fields like cluster_id, status, and computed fields cannot be modified. ## OpenAPI ````yaml patch /v1/clusters/{cluster_identifier} openapi: 3.1.0 info: title: Mixpeek API description: >- This is the Mixpeek API, providing access to various endpoints for data processing and retrieval. termsOfService: https://mixpeek.com/terms contact: name: Mixpeek Support url: https://mixpeek.com/contact email: info@mixpeek.com version: '0.82' servers: - url: https://api.mixpeek.com description: Production security: [] paths: /v1/clusters/{cluster_identifier}: patch: tags: - Clusters summary: Partially Update Cluster description: |- This endpoint partially updates a cluster (PATCH operation). Only provided fields will be updated. At minimum, metadata can always be updated. Immutable fields like cluster_id, status, and computed fields cannot be modified. operationId: patch_cluster_v1_clusters__cluster_identifier__patch parameters: - name: cluster_identifier in: path required: true schema: type: string description: Cluster ID or name title: Cluster Identifier description: Cluster ID or name requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/PatchClusterRequest' responses: '200': description: Successful Response content: application/json: schema: $ref: '#/components/schemas/ClusterModel' '400': description: Bad Request content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '401': description: Unauthorized content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '403': description: Forbidden content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '404': description: Not Found content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '422': description: Validation Error content: application/json: schema: $ref: '#/components/schemas/HTTPValidationError' '500': description: Internal Server Error content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' components: schemas: PatchClusterRequest: properties: cluster_name: anyOf: - type: string - type: 'null' title: Cluster Name description: Updated name for the cluster description: anyOf: - type: string - type: 'null' title: Description description: Updated description for the cluster metadata: anyOf: - additionalProperties: true type: object - type: 'null' title: Metadata description: Updated metadata for the cluster llm_labeling: anyOf: - $ref: '#/components/schemas/LLMLabeling-Input' - type: 'null' description: >- Updated LLM labeling configuration. Takes effect on the next `POST /v1/clusters/{id}/execute` ā€” use this to correct a null `labeling_inputs` mapping that produced schema-metadata labels, without re-embedding or re-running HDBSCAN. filters: anyOf: - additionalProperties: true type: object - type: 'null' title: Filters description: >- Updated pre-filter for clustering input documents. Overrides the cluster's stored filter on subsequent execute calls. face_cluster_merge: anyOf: - $ref: '#/components/schemas/FaceClusterMergeConfig' - type: 'null' description: >- Updated post-HDBSCAN face-identity merge configuration. Takes effect on the next `POST /v1/clusters/{id}/execute`. Pass an object with enabled=false to turn the merge pass off without removing the config; pass null in the patch to leave the stored value untouched. sample_size: anyOf: - type: integer maximum: 1000000 - type: 'null' title: Sample Size description: >- Updated per-execution document cap. Takes effect on the next `POST /v1/clusters/{id}/execute`. Omit to leave the stored value untouched; set to an integer to change it. KMeans supports up to 1M; O(NĀ²) algorithms are capped at 100K by the engine. algorithm_params: anyOf: - additionalProperties: true type: object - type: 'null' title: Algorithm Params description: >- Updated algorithm parameters (e.g. min_cluster_size, min_samples for HDBSCAN). Takes effect on the next `POST /v1/clusters/{id}/execute`. type: object title: PatchClusterRequest description: Request model for partially updating a cluster (PATCH operation). ClusterModel: properties: collection_ids: anyOf: - items: type: string type: array minItems: 1 - type: 'null' title: Collection Ids description: Collections to cluster together cluster_name: anyOf: - type: string - type: 'null' title: Cluster Name description: Optional human-friendly name for the clustering job cluster_type: $ref: '#/components/schemas/ClusterType' description: Vector or attribute clustering default: vector vector_config: anyOf: - $ref: '#/components/schemas/VectorBasedConfig-Output' - type: 'null' description: Required when cluster_type is 'vector' attribute_config: anyOf: - $ref: '#/components/schemas/AttributeBasedConfig' - type: 'null' description: Required when cluster_type is 'attribute' filters: anyOf: - $ref: '#/components/schemas/LogicalOperator-Output' - type: 'null' description: >- Optional filters to pre-filter documents before clustering (same format as list documents). Applied during Qdrant scroll before parquet export. Useful for clustering subsets like: status='active', category='electronics', etc. llm_labeling: anyOf: - $ref: '#/components/schemas/LLMLabeling-Output' - type: 'null' description: >- Optional configuration for LLM-based cluster labeling. When provided with enabled=True, clusters will have semantic labels generated by LLM instead of generic labels like 'Cluster 0'. When not provided or enabled=False, uses fallback labels. enrich_source_collection: type: boolean title: Enrich Source Collection description: >- If True, cluster results are written back to source collection(s) in-place instead of creating new output collections. Documents will be enriched with cluster_id, cluster_label, distance_to_centroid, and optionally other metadata. Similar to taxonomy enrichment pattern. default: false source_enrichment_config: anyOf: - $ref: '#/components/schemas/SourceEnrichmentConfig' - type: 'null' description: >- Configuration for source collection enrichment (only used if enrich_source_collection=True). Controls which fields are added to source documents and field naming conventions. auto_execute_on_batch: type: boolean title: Auto Execute On Batch description: >- Automatically execute this cluster whenever a batch completes on any of its input collections. When True, a ClusterApplicationConfig entry is added to each input collection's cluster_applications field at creation time. The cluster will then auto-trigger after each batch completion (subject to cooldown and document threshold). When False (default), the cluster must be executed manually via the API. default: false auto_execute_min_documents: anyOf: - type: integer - type: 'null' title: Auto Execute Min Documents description: >- Minimum number of documents required before auto-executing cluster. Only used when auto_execute_on_batch=True. If the collection has fewer documents than this threshold, clustering is skipped. auto_execute_cooldown_seconds: type: integer title: Auto Execute Cooldown Seconds description: >- Minimum time (in seconds) between automatic cluster executions. Only used when auto_execute_on_batch=True. Default: 3600 (1 hour). default: 3600 cluster_id: type: string title: Cluster Id description: Unique cluster identifier parquet_path: anyOf: - type: string - type: 'null' title: Parquet Path description: S3 path to parquet files with cluster data members_key: anyOf: - type: string - type: 'null' title: Members Key description: S3 key to members.parquet (if saved) num_clusters: anyOf: - type: integer - type: 'null' title: Num Clusters description: Number of clusters found cluster_stats: anyOf: - $ref: '#/components/schemas/ClusterStats' - type: 'null' description: Clustering quality metrics status: $ref: '#/components/schemas/TaskStatusEnum' description: Clustering job status default: PENDING task_id: anyOf: - type: string - type: 'null' title: Task Id description: Associated task ID for clustering job last_run_id: anyOf: - type: string - type: 'null' title: Last Run Id description: >- Run ID of the most recent successful clustering execution. Used to retrieve execution results. created_at: type: string format: date-time title: Created At description: When the cluster was created updated_at: type: string format: date-time title: Updated At description: When the cluster was last updated metadata: additionalProperties: true type: object title: Metadata description: Additional user-defined metadata for the cluster type: object title: ClusterModel description: Cluster metadata stored in MongoDB. examples: - cluster_name: products_clip_hdbscan cluster_type: vector collection_ids: - col_products_v1 - col_products_v2 llm_labeling: enabled: true labeling_inputs: input_mappings: - input_key: text path: description source_type: payload model_name: gpt-4o-mini-2024-07-18 provider: openai vector_config: clustering_method: hdbscan feature_uri: mixpeek://clip_vit_l_14@v1/embedding hdbscan_parameters: min_cluster_size: 10 min_samples: 5 sample_size: 5000 - cluster_name: video_multimodal_clustering cluster_type: vector collection_ids: - col_videos_v1 llm_labeling: enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: title source_type: payload - input_key: video_url path: video_url source_type: payload model_name: gemini-2.5-flash-lite provider: google vector_config: clustering_method: hdbscan feature_uri: mixpeek://clip_vit_l_14@v1/embedding hdbscan_parameters: min_cluster_size: 5 min_samples: 3 sample_size: 3000 - cluster_name: image_multimodal_clustering cluster_type: vector collection_ids: - col_images_v1 llm_labeling: enabled: true labeling_inputs: input_mappings: - input_key: text path: caption source_type: payload - input_key: image_url path: image_url source_type: payload model_name: gemini-2.5-flash-lite provider: google vector_config: clustering_method: kmeans feature_uri: mixpeek://clip_vit_l_14@v1/embedding kmeans_parameters: n_clusters: 10 sample_size: 2000 - attribute_config: attributes: - status hierarchical_grouping: true cluster_name: orders_group_by_status cluster_type: attribute collection_ids: - col_orders_v1 - col_orders_v2 - col_orders_v3 ErrorResponse: properties: success: type: boolean title: Success description: Always false for error responses default: false status: type: integer title: Status description: HTTP status code for this error error: $ref: '#/components/schemas/ErrorDetail' description: Error details payload type: object required: - status - error title: ErrorResponse description: Error response model. examples: - error: details: id: ns_123 resource: namespace message: Namespace not found type: NotFoundError status: 404 success: false HTTPValidationError: properties: detail: items: $ref: '#/components/schemas/ValidationError' type: array title: Detail type: object title: HTTPValidationError LLMLabeling-Input: properties: enabled: type: boolean title: Enabled description: >- Whether to generate labels for clusters using LLM. When enabled, clusters will have semantic labels like 'High-Performance Laptops' instead of generic labels like 'Cluster 0'. default: false labeling_inputs: anyOf: - $ref: '#/components/schemas/LLMLabelingInput-Input' - type: 'null' description: >- Input configuration for LLM labeling. Supports flexible input mappings for multimodal inputs (text, images, videos, audio). Use input_mappings for advanced multimodal labeling with providers like Gemini. If not provided (null/undefined), the full document payload is serialized as JSON and sent to the LLM ā€” WARNING: in practice this produces schema-metadata labels (e.g. 'Mid-Timeline Video Events' with keywords like ['start_time', 'end_time']) because the LLM describes whatever field names it sees. For meaningful labels, set input_mappings that point at semantic fields (title, description, thumbnail_url, document_blobs[].url for image/video content) rather than relying on the default. provider: anyOf: - $ref: '#/components/schemas/LLMProvider' - type: 'null' description: |- LLM provider to use for labeling. Supported providers: - openai: GPT models (GPT-4o, GPT-4o-mini, GPT-4.1, O3-mini) - google: Gemini models (Gemini 2.5 Flash, Gemini 1.5 Flash) - anthropic: Claude models (Claude 3.5 Sonnet, Claude 3.5 Haiku) If not specified, automatically inferred from model_name. examples: - openai - google - anthropic model_name: anyOf: - $ref: '#/components/schemas/OpenAIModel' - $ref: '#/components/schemas/GoogleModel' - $ref: '#/components/schemas/AnthropicModel' - type: 'null' title: Model Name description: >- REQUIRED when enabled=True. Specific LLM model to use for cluster labeling. All models are defined as enums for type safety. OpenAI Models (provider='openai'): - gpt-4o-2024-08-06: Highest quality, best for production ($2.50/$10 per 1M tokens) - gpt-4o-mini-2024-07-18: Cost-effective, recommended for most use cases ($0.15/$0.60 per 1M tokens) - gpt-4.1-2025-04-14: Latest model, future-proofed - gpt-4.1-mini-2025-04-14: Latest cost-optimized model - o3-mini-2025-01-31: Advanced reasoning, best for complex clustering Google Models (provider='google'): - gemini-2.5-flash-lite: Fastest, latest multimodal model, recommended ($0.15/$0.60 per 1M tokens) Anthropic Models (provider='anthropic'): - claude-3-5-sonnet-20241022: Best reasoning, 200K context ($3/$15 per 1M tokens) - claude-3-5-haiku-20241022: Fast, cost-effective ($0.25/$1.25 per 1M tokens) Recommendation: - Use gemini-2.5-flash-lite (DEFAULT) - multimodal support - Use gpt-4o-mini-2024-07-18 for OpenAI compatibility - Use gpt-4o-2024-08-06 for highest quality when cost is not a concern examples: - gemini-2.5-flash-lite - gemini-2.5-pro - gpt-4o-mini-2024-07-18 - gpt-4o-2024-08-06 - claude-sonnet-4-5-20250929 - claude-haiku-4-5-20251001 - gpt-4.1-2025-04-14 - gpt-4.1-mini-2025-04-14 - o3-mini-2025-01-31 include_summary: type: boolean title: Include Summary description: Whether to generate cluster summaries default: true include_keywords: type: boolean title: Include Keywords description: Whether to extract keywords for clusters default: true max_samples_per_cluster: anyOf: - type: integer maximum: 20 minimum: 1 - type: 'null' title: Max Samples Per Cluster description: >- Maximum representative documents to send to LLM per cluster for semantic analysis. When null (default), automatically scales based on cluster size and spread ā€” smaller/tighter clusters get fewer samples, larger/sparser clusters get more (range 3-20). Set explicitly to override with a fixed value. sample_text_max_length: type: integer maximum: 500 minimum: 50 title: Sample Text Max Length description: Maximum characters per document sample text default: 200 use_embedding_dedup: type: boolean title: Use Embedding Dedup description: >- Enable embedding-based label deduplication to prevent near-duplicate labels (requires sentence-transformers) default: true embedding_similarity_threshold: type: number maximum: 1 minimum: 0.5 title: Embedding Similarity Threshold description: >- Cosine similarity threshold for duplicate label detection (labels above this are considered duplicates) default: 0.8 cache_ttl_seconds: type: integer maximum: 2592000 minimum: 0 title: Cache Ttl Seconds description: >- Time-to-live for cached labels in seconds. Labels for clusters with identical representative documents will be reused within this TTL window, reducing LLM API costs. Default: 604800 (7 days). Set to 0 to disable caching. default: 604800 custom_prompt: anyOf: - type: string - type: 'null' title: Custom Prompt description: >- OPTIONAL. Custom prompt template for LLM labeling. NOT REQUIRED - uses default discriminative prompt if not provided. When provided, completely replaces the default prompt. Your custom prompt receives cluster information but you must format it yourself. Use when: - Need domain-specific labeling (e.g., medical, legal, technical) - Want different label format (e.g., emoji labels, code names) - Require specific output structure - Have custom business logic for categorization Default prompt includes: cluster document samples, forbidden labels for uniqueness, and JSON response format. See engine/clusters/labeling/prompts.py for reference. Example: 'Analyze these product clusters and generate SHORT category names (2-3 words max) focusing on product type and price range. Return JSON: [{"cluster_id": "cl_0", "label": "..."}]' examples: - >- Analyze these document clusters and generate technical labels (2-3 words). Focus on programming languages and frameworks mentioned. Return JSON: [{'cluster_id': 'cl_0', 'label': '...', 'keywords': [...]}] - >- Generate emoji-based labels for these clusters. Use 1-2 emojis that represent the main theme. Return JSON: [{'cluster_id': 'cl_0', 'label': 'šŸš€ Tech Innovation'}] - null response_shape: anyOf: - type: string - additionalProperties: true type: object - type: 'null' title: Response Shape description: > OPTIONAL. Define custom structured output for LLM labeling. NOT REQUIRED - uses default structure (label, summary, keywords) if not provided. When provided, LLM output will match this structure and be stored in cluster documents. Two modes supported: 1. Natural language prompt (string): Describe desired output in plain English - Service automatically infers JSON schema from your description - Example: 'Extract cluster category, confidence score (0-1), and top 3 representative terms' - Auto-generates schema with appropriate types (string, number, array, etc.) 2. Explicit JSON schema (dict): Provide complete JSON schema for output structure - Full control over output structure, types, and constraints - Example: {'type': 'object', 'properties': {'category': {'type': 'string'}, ...}} Use when: - Need custom metadata fields (confidence scores, sentiment, complexity) - Want domain-specific structure (taxonomy hierarchies, entity extractions) - Require specific data types (arrays, nested objects, enums) - Have downstream schema requirements Output fields are automatically added to cluster collection schema and stored in metadata. Default behavior (if not provided): label (string), summary (string), keywords (array of strings) examples: - >- Extract cluster category, confidence score between 0 and 1, and top 3 representative keywords - >- Generate cluster theme, sentiment (positive/negative/neutral), and list of key entities - properties: category: description: Main category type: string subcategory: description: Subcategory if applicable type: string confidence: maximum: 1 minimum: 0 type: number keywords: items: type: string maxItems: 5 type: array required: - category - confidence type: object - null parameters: additionalProperties: true type: object title: Parameters description: >- Provider-specific parameters forwarded to the LLM service. For OpenAI: temperature, max_tokens, top_p, json_output, etc. For Google: temperature, top_k, max_output_tokens, json_output, etc. type: object title: LLMLabeling description: |- Configuration for LLM-based cluster labeling. Supports multiple LLM providers with comprehensive model selection: - OpenAI: GPT-4o, GPT-4o-mini, GPT-4.1, O3-mini (best for quality) - Google: Gemini 2.5 Flash, Gemini 1.5 Flash (best for speed and cost) - Anthropic: Claude 3.5 Sonnet, Claude 3.5 Haiku (best for reasoning) All models are defined as enums and validated at API level. examples: - description: Text-only labeling with multiple fields enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: title path: title source_type: payload - input_key: description path: description source_type: payload - input_key: text path: text source_type: payload model_name: gpt-4o-mini-2024-07-18 provider: openai - description: Multimodal labeling with images (Gemini) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: headline source_type: payload - input_key: image_url path: thumbnail_url source_type: payload model_name: gemini-2.5-flash-lite provider: google - description: Multimodal labeling with video (Gemini) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: description source_type: payload - input_key: video_url path: video_url source_type: payload model_name: gemini-2.5-flash-lite provider: google - description: OpenAI GPT-4o (highest quality, text-only) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: gpt-4o-2024-08-06 provider: openai - description: Anthropic Claude 3.5 Sonnet (best reasoning) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: claude-sonnet-4-5-20250929 provider: anthropic - description: 'Minimal configuration (uses defaults: text field from payload)' enabled: true - custom_prompt: >- You are a medical document classifier. Analyze the following patient record clusters and generate HIPAA-compliant category labels (2-3 words) that describe the medical condition or treatment type. DO NOT include patient names or identifiers. Return JSON array: [{"cluster_id": "cl_0", "label": "...", "keywords": [...]}] description: Custom prompt for domain-specific labeling enabled: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: gpt-4o-mini-2024-07-18 provider: openai FaceClusterMergeConfig: properties: enabled: type: boolean title: Enabled description: >- Run the merge pass. Set False to disable without removing the config. default: true centroid_cosine_threshold: type: number maximum: 1 minimum: 0 title: Centroid Cosine Threshold description: Minimum centroid cosine similarity for a candidate merge. default: 0.55 bbox_iou_threshold: type: number maximum: 1 minimum: 0 title: Bbox Iou Threshold description: Minimum bbox IoU on overlapping frames to satisfy the spatial half. default: 0.4 scene_jaccard_threshold: type: number maximum: 1 minimum: 0 title: Scene Jaccard Threshold description: >- Minimum Jaccard similarity of scene-id sets to satisfy the spatial half. default: 0.3 bbox_field: type: string title: Bbox Field description: >- Document-payload field holding the face bbox (list/tuple of 4 floats). default: bbox frame_field: type: string title: Frame Field description: >- Document-payload field holding the frame identifier used to pair bboxes. default: frame_number scene_field: type: string title: Scene Field description: >- Document-payload field holding the scene identifier used for Jaccard. default: scene_id type: object title: FaceClusterMergeConfig description: |- Configuration for the post-HDBSCAN face-identity merge pass. Enables an agglomerative merge after HDBSCAN labels are assigned but before centroid calculation. Two clusters merge when the centroid cosine meets the cosine threshold AND at least one of the spatial signals (bbox IoU on overlapping frames, scene Jaccard) also clears its threshold. Defaults target ArcFace 512d face embeddings at the brand-corpus scale (~10^4 faces, ~10^2 true identities). ClusterType: type: string enum: - vector - attribute title: ClusterType description: >- Type of clustering to perform. Determines the clustering approach: - vector: Cluster documents by embedding similarity (semantic clustering) - attribute: Cluster documents by metadata attributes (business logic clustering) Use Cases: vector: - Group semantically similar content - Find content with similar meaning - Organize by topic/theme - Requires vector embeddings attribute: - Group by business attributes (category, brand, status, etc.) - Organize by explicit metadata - Create hierarchical groupings - No embeddings required VectorBasedConfig-Output: properties: feature_uri: anyOf: - type: string - type: 'null' title: Feature Uri description: >- DEPRECATED: Use feature_uris instead. Canonical feature URI for the vector embedding to cluster. Format: 'mixpeek://{extractor}@{version}/{output}'. For multi-feature clustering, use feature_uris (plural) instead. examples: - mixpeek://multimodal_extractor@v1/vertex_multimodal_embedding - mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1 feature_uris: anyOf: - items: type: string type: array minItems: 1 - type: 'null' title: Feature Uris description: >- RECOMMENDED. List of feature URIs to cluster. Format: 'mixpeek://{extractor}@{version}/{output}'. For single-feature clustering, provide a list with one element. For multi-feature clustering, provide multiple feature URIs. Each feature must exist in all input collections. examples: - - mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1 - - mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1 - mixpeek://image_extractor@v1/embedding clustering_method: $ref: '#/components/schemas/ClusteringAlgorithm' description: Clustering algorithm to use sample_size: anyOf: - type: integer maximum: 1000000 - type: 'null' title: Sample Size description: >- Number of samples to use for clustering. If not set, defaults are applied per algorithm to prevent out-of-memory: HDBSCAN/DBSCAN/OPTICS: 50,000 (O(NĀ²) memory), Spectral/Agglomerative: 50,000, KMeans/GaussianMixture/MeanShift: 100,000. KMeans/GMM hard max: 1,000,000; O(NĀ²) algorithms: 100,000. kmeans_parameters: anyOf: - $ref: '#/components/schemas/KMeansParams' - additionalProperties: true type: object - type: 'null' title: Kmeans Parameters description: Parameters for K-means clustering (deprecated, use algorithm_params) dbscan_parameters: anyOf: - $ref: '#/components/schemas/DBSCANParams' - additionalProperties: true type: object - type: 'null' title: Dbscan Parameters description: Parameters for DBSCAN clustering (deprecated, use algorithm_params) hdbscan_parameters: anyOf: - $ref: '#/components/schemas/HDBSCANParams' - additionalProperties: true type: object - type: 'null' title: Hdbscan Parameters description: Parameters for HDBSCAN clustering (deprecated, use algorithm_params) algorithm_params: anyOf: - $ref: '#/components/schemas/KMeansParams' - $ref: '#/components/schemas/DBSCANParams' - $ref: '#/components/schemas/HDBSCANParams' - $ref: '#/components/schemas/AgglomerativeParams' - $ref: '#/components/schemas/SpectralParams' - $ref: '#/components/schemas/GaussianMixtureParams' - $ref: '#/components/schemas/MeanShiftParams' - $ref: '#/components/schemas/OPTICSParams' - $ref: '#/components/schemas/LeidenParams' - additionalProperties: true type: object - type: 'null' title: Algorithm Params description: Algorithm-specific parameters multi_feature_strategy: type: string enum: - concatenate - independent - weighted title: Multi Feature Strategy description: |- Strategy for handling multiple feature vectors: - concatenate: Combine embeddings into one vector, single clustering - independent: Run separate clustering per feature - weighted: Learn optimal feature weights default: concatenate normalize_features: type: boolean title: Normalize Features description: >- Apply L2 normalization to each feature block before concatenation. Prevents feature dominance when combining different modalities. Only applies when multi_feature_strategy='concatenate'. default: true feature_weights: anyOf: - additionalProperties: type: number type: object - type: 'null' title: Feature Weights description: >- Optional per-feature weights (0.0-1.0) for concatenation strategy. Keys are feature URIs, values are weights. Example: {'mixpeek://text@v1/emb': 0.7, 'mixpeek://image@v1/emb': 0.3}. Defaults to equal weights (1.0) if not specified. Only applies when multi_feature_strategy='concatenate'. If multi_feature_strategy='weighted' and this is None, weights are learned automatically using weight_learning_config. examples: - mixpeek://image_extractor@v1/embedding: 0.3 mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1: 0.7 weight_learning_config: anyOf: - $ref: '#/components/schemas/WeightLearningConfig' - type: 'null' description: >- Configuration for automatic feature weight learning. Only used when multi_feature_strategy='weighted' and feature_weights is None. If feature_weights is provided, manual weights are used instead of learning. If this is None when learning is needed, default WeightLearningConfig is used. examples: - max_iterations: 20 method: bayesian metric: silhouette sample_size: 5000 - max_iterations: 5 method: grid_search metric: silhouette output_strategy: type: string enum: - single - per_feature title: Output Strategy description: >- Output collection creation strategy: - single: Create one collection with all feature vectors - per_feature: Create separate collections for each feature (for hierarchical taxonomies) default: single effective_feature_method: type: string enum: - mean - median - medoid title: Effective Feature Method description: |- Method for calculating cluster centroids: - mean: Average of all vectors in cluster - median: Median vector (robust to outliers) - medoid: Actual cluster member closest to centroid default: mean enrich_source: type: boolean title: Enrich Source description: Whether to enrich source documents with cluster_id default: false face_cluster_merge: anyOf: - $ref: '#/components/schemas/FaceClusterMergeConfig' - type: 'null' description: >- Optional post-HDBSCAN agglomerative merge pass for face identity clusters. HDBSCAN over ArcFace 512d face embeddings tends to over-split the same person across pose/lighting variants. When provided with enabled=true, two clusters merge if their centroid cosine meets the threshold AND one of the spatial signals (bbox IoU on overlapping frames, scene-id Jaccard) clears its threshold. Leave null (default) for non-face clusterings ā€” existing behavior is preserved. preprocessing_steps: anyOf: - items: oneOf: - $ref: '#/components/schemas/TSNEParams' - $ref: '#/components/schemas/UMAPParams' - $ref: '#/components/schemas/WhiteningParams' - $ref: '#/components/schemas/NoReduction' discriminator: propertyName: method mapping: none: $ref: '#/components/schemas/NoReduction' tsne: $ref: '#/components/schemas/TSNEParams' umap: $ref: '#/components/schemas/UMAPParams' whitening: $ref: '#/components/schemas/WhiteningParams' type: array - type: 'null' title: Preprocessing Steps description: >- Ordered list of preprocessing steps applied before clustering. Steps execute in order. Common patterns: - [whitening, umap]: Decorrelate then reduce ā€” best for high-dimensional embeddings - [umap]: UMAP pre-reduction only (defaults: 50D, cosine, n_neighbors=30) - [whitening]: Whitening only ā€” improves density-based clustering without dimension reduction If set, overrides any default dimensionality reduction. examples: - - method: whitening regularization: 0.00001 - method: umap min_dist: 0 n_components: 50 n_neighbors: 30 - - method: umap n_components: 50 hierarchical: type: boolean title: Hierarchical description: >- Enable recursive sub-clustering. After initial clustering, each cluster with enough members is sub-divided using UMAP+HDBSCAN recursively. Produces hierarchical cluster IDs (e.g., cl_0_sub_1_sub_0). default: false max_hierarchy_depth: type: integer maximum: 5 minimum: 1 title: Max Hierarchy Depth description: Maximum recursion depth for hierarchical sub-clustering. default: 3 vis_n_components: type: integer enum: - 2 - 3 title: Vis N Components description: >- Number of dimensions for visualization coordinates (2 or 3). When 3, the z coordinate is populated for depth/size-based rendering. Stored on the cluster and used as the default for all executions. default: 2 type: object required: - clustering_method title: VectorBasedConfig description: >- Configuration for vector-based clustering. Use canonical feature URIs to specify which vector embeddings to cluster. Feature URIs follow the format: mixpeek://{extractor}@{version}/{output} Supports both single and multi-feature clustering: - Single feature: Provide one feature_uri for standard clustering - Multi-feature: Provide multiple feature_uris for hybrid clustering Examples: Single feature: { "feature_uri": "mixpeek://multimodal_extractor@v1/vertex_multimodal_embedding", "clustering_method": "hdbscan", "sample_size": 1000 } Multi-feature: { "feature_uris": [ "mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1", "mixpeek://image_extractor@v1/embedding" ], "clustering_method": "hdbscan", "multi_feature_strategy": "concatenate" } examples: - algorithm_params: min_cluster_size: 10 min_samples: 5 clustering_method: hdbscan description: HDBSCAN clustering with multimodal embeddings feature_uri: mixpeek://multimodal_extractor@v1/vertex_multimodal_embedding sample_size: 1000 - clustering_method: hdbscan description: Face-identity clustering with post-HDBSCAN merge face_cluster_merge: bbox_iou_threshold: 0.4 centroid_cosine_threshold: 0.55 enabled: true scene_jaccard_threshold: 0.3 feature_uris: - mixpeek://face_identity_extractor@v1/face_embedding - algorithm_params: n_clusters: 10 clustering_method: kmeans description: K-means clustering with text embeddings feature_uri: mixpeek://text_extractor@v1/multilingual_e5_large_instruct_v1 - algorithm_params: eps: 0.5 min_samples: 5 clustering_method: dbscan description: DBSCAN clustering with CLIP image embeddings feature_uri: mixpeek://clip@v1/image_embedding AttributeBasedConfig: properties: attributes: items: type: string type: array minItems: 1 title: Attributes description: >- List of attribute field names to use for clustering. Documents will be grouped by unique combinations of these attribute values. Supports dot-notation for nested fields (e.g., 'metadata.category'). Order matters for hierarchical grouping: first attribute is top-level, subsequent are nested. examples: - - category - - category - brand - - status - priority - - metadata.author - metadata.topic hierarchical_grouping: type: boolean title: Hierarchical Grouping description: >- Whether to create hierarchical clusters based on attribute order. When True: Creates parent clusters for each unique value of the first attribute, then child clusters for subsequent attributes within each parent. When False: Creates flat clusters for each unique combination of all attributes. Example with ['category', 'brand']: hierarchical=True ā†’ 'Electronics' (parent) ā†’ 'Apple', 'Samsung' (children). hierarchical=False ā†’ 'Electronics_Apple', 'Electronics_Samsung' (flat). default: false aggregation_method: anyOf: - type: string - type: 'null' title: Aggregation Method description: >- Method for aggregating attribute values when creating cluster centroids. Options: 'most_frequent' (default), 'first', 'last'. Most use cases should use the default. examples: - most_frequent - first - last type: object required: - attributes title: AttributeBasedConfig description: >- Configuration for attribute-based clustering. Attribute-based clustering groups documents by metadata attributes (e.g., category, brand, status) instead of vector similarity. This is useful for organizing content by business logic rather than semantic similarity. Examples: - Group products by category and brand - Organize orders by status and priority - Cluster content by author and topic examples: - attributes: - category description: Simple category clustering hierarchical_grouping: false - attributes: - category - brand description: Hierarchical category ā†’ brand clustering hierarchical_grouping: true - aggregation_method: most_frequent attributes: - status - priority description: Order status and priority (flat) hierarchical_grouping: false LogicalOperator-Output: properties: AND: anyOf: - items: anyOf: - $ref: '#/components/schemas/LogicalOperator-Output' - $ref: '#/components/schemas/FilterCondition' type: array - type: 'null' title: And description: Logical AND operation - all conditions must be true example: - field: name operator: eq value: John - field: age operator: gte value: 30 OR: anyOf: - items: anyOf: - $ref: '#/components/schemas/LogicalOperator-Output' - $ref: '#/components/schemas/FilterCondition' type: array - type: 'null' title: Or description: Logical OR operation - at least one condition must be true example: - field: status operator: eq value: active - field: role operator: eq value: admin NOT: anyOf: - items: anyOf: - $ref: '#/components/schemas/LogicalOperator-Output' - $ref: '#/components/schemas/FilterCondition' type: array - type: 'null' title: Not description: Logical NOT operation - all conditions must be false example: - field: department operator: eq value: HR - field: location operator: eq value: remote case_sensitive: anyOf: - type: boolean - type: 'null' title: Case Sensitive description: Whether to perform case-sensitive matching default: false example: true additionalProperties: true type: object title: LogicalOperator description: >- Represents a logical operation (AND, OR, NOT) on filter conditions. Allows nesting with a defined depth limit. Also supports shorthand syntax where field names can be passed directly as key-value pairs for equality filtering (e.g., {"metadata.title": "value"}). LLMLabeling-Output: properties: enabled: type: boolean title: Enabled description: >- Whether to generate labels for clusters using LLM. When enabled, clusters will have semantic labels like 'High-Performance Laptops' instead of generic labels like 'Cluster 0'. default: false labeling_inputs: anyOf: - $ref: '#/components/schemas/LLMLabelingInput-Output' - type: 'null' description: >- Input configuration for LLM labeling. Supports flexible input mappings for multimodal inputs (text, images, videos, audio). Use input_mappings for advanced multimodal labeling with providers like Gemini. If not provided (null/undefined), the full document payload is serialized as JSON and sent to the LLM ā€” WARNING: in practice this produces schema-metadata labels (e.g. 'Mid-Timeline Video Events' with keywords like ['start_time', 'end_time']) because the LLM describes whatever field names it sees. For meaningful labels, set input_mappings that point at semantic fields (title, description, thumbnail_url, document_blobs[].url for image/video content) rather than relying on the default. provider: anyOf: - $ref: '#/components/schemas/LLMProvider' - type: 'null' description: |- LLM provider to use for labeling. Supported providers: - openai: GPT models (GPT-4o, GPT-4o-mini, GPT-4.1, O3-mini) - google: Gemini models (Gemini 2.5 Flash, Gemini 1.5 Flash) - anthropic: Claude models (Claude 3.5 Sonnet, Claude 3.5 Haiku) If not specified, automatically inferred from model_name. examples: - openai - google - anthropic model_name: anyOf: - $ref: '#/components/schemas/OpenAIModel' - $ref: '#/components/schemas/GoogleModel' - $ref: '#/components/schemas/AnthropicModel' - type: 'null' title: Model Name description: >- REQUIRED when enabled=True. Specific LLM model to use for cluster labeling. All models are defined as enums for type safety. OpenAI Models (provider='openai'): - gpt-4o-2024-08-06: Highest quality, best for production ($2.50/$10 per 1M tokens) - gpt-4o-mini-2024-07-18: Cost-effective, recommended for most use cases ($0.15/$0.60 per 1M tokens) - gpt-4.1-2025-04-14: Latest model, future-proofed - gpt-4.1-mini-2025-04-14: Latest cost-optimized model - o3-mini-2025-01-31: Advanced reasoning, best for complex clustering Google Models (provider='google'): - gemini-2.5-flash-lite: Fastest, latest multimodal model, recommended ($0.15/$0.60 per 1M tokens) Anthropic Models (provider='anthropic'): - claude-3-5-sonnet-20241022: Best reasoning, 200K context ($3/$15 per 1M tokens) - claude-3-5-haiku-20241022: Fast, cost-effective ($0.25/$1.25 per 1M tokens) Recommendation: - Use gemini-2.5-flash-lite (DEFAULT) - multimodal support - Use gpt-4o-mini-2024-07-18 for OpenAI compatibility - Use gpt-4o-2024-08-06 for highest quality when cost is not a concern examples: - gemini-2.5-flash-lite - gemini-2.5-pro - gpt-4o-mini-2024-07-18 - gpt-4o-2024-08-06 - claude-sonnet-4-5-20250929 - claude-haiku-4-5-20251001 - gpt-4.1-2025-04-14 - gpt-4.1-mini-2025-04-14 - o3-mini-2025-01-31 include_summary: type: boolean title: Include Summary description: Whether to generate cluster summaries default: true include_keywords: type: boolean title: Include Keywords description: Whether to extract keywords for clusters default: true max_samples_per_cluster: anyOf: - type: integer maximum: 20 minimum: 1 - type: 'null' title: Max Samples Per Cluster description: >- Maximum representative documents to send to LLM per cluster for semantic analysis. When null (default), automatically scales based on cluster size and spread ā€” smaller/tighter clusters get fewer samples, larger/sparser clusters get more (range 3-20). Set explicitly to override with a fixed value. sample_text_max_length: type: integer maximum: 500 minimum: 50 title: Sample Text Max Length description: Maximum characters per document sample text default: 200 use_embedding_dedup: type: boolean title: Use Embedding Dedup description: >- Enable embedding-based label deduplication to prevent near-duplicate labels (requires sentence-transformers) default: true embedding_similarity_threshold: type: number maximum: 1 minimum: 0.5 title: Embedding Similarity Threshold description: >- Cosine similarity threshold for duplicate label detection (labels above this are considered duplicates) default: 0.8 cache_ttl_seconds: type: integer maximum: 2592000 minimum: 0 title: Cache Ttl Seconds description: >- Time-to-live for cached labels in seconds. Labels for clusters with identical representative documents will be reused within this TTL window, reducing LLM API costs. Default: 604800 (7 days). Set to 0 to disable caching. default: 604800 custom_prompt: anyOf: - type: string - type: 'null' title: Custom Prompt description: >- OPTIONAL. Custom prompt template for LLM labeling. NOT REQUIRED - uses default discriminative prompt if not provided. When provided, completely replaces the default prompt. Your custom prompt receives cluster information but you must format it yourself. Use when: - Need domain-specific labeling (e.g., medical, legal, technical) - Want different label format (e.g., emoji labels, code names) - Require specific output structure - Have custom business logic for categorization Default prompt includes: cluster document samples, forbidden labels for uniqueness, and JSON response format. See engine/clusters/labeling/prompts.py for reference. Example: 'Analyze these product clusters and generate SHORT category names (2-3 words max) focusing on product type and price range. Return JSON: [{"cluster_id": "cl_0", "label": "..."}]' examples: - >- Analyze these document clusters and generate technical labels (2-3 words). Focus on programming languages and frameworks mentioned. Return JSON: [{'cluster_id': 'cl_0', 'label': '...', 'keywords': [...]}] - >- Generate emoji-based labels for these clusters. Use 1-2 emojis that represent the main theme. Return JSON: [{'cluster_id': 'cl_0', 'label': 'šŸš€ Tech Innovation'}] - null response_shape: anyOf: - type: string - additionalProperties: true type: object - type: 'null' title: Response Shape description: > OPTIONAL. Define custom structured output for LLM labeling. NOT REQUIRED - uses default structure (label, summary, keywords) if not provided. When provided, LLM output will match this structure and be stored in cluster documents. Two modes supported: 1. Natural language prompt (string): Describe desired output in plain English - Service automatically infers JSON schema from your description - Example: 'Extract cluster category, confidence score (0-1), and top 3 representative terms' - Auto-generates schema with appropriate types (string, number, array, etc.) 2. Explicit JSON schema (dict): Provide complete JSON schema for output structure - Full control over output structure, types, and constraints - Example: {'type': 'object', 'properties': {'category': {'type': 'string'}, ...}} Use when: - Need custom metadata fields (confidence scores, sentiment, complexity) - Want domain-specific structure (taxonomy hierarchies, entity extractions) - Require specific data types (arrays, nested objects, enums) - Have downstream schema requirements Output fields are automatically added to cluster collection schema and stored in metadata. Default behavior (if not provided): label (string), summary (string), keywords (array of strings) examples: - >- Extract cluster category, confidence score between 0 and 1, and top 3 representative keywords - >- Generate cluster theme, sentiment (positive/negative/neutral), and list of key entities - properties: category: description: Main category type: string subcategory: description: Subcategory if applicable type: string confidence: maximum: 1 minimum: 0 type: number keywords: items: type: string maxItems: 5 type: array required: - category - confidence type: object - null parameters: additionalProperties: true type: object title: Parameters description: >- Provider-specific parameters forwarded to the LLM service. For OpenAI: temperature, max_tokens, top_p, json_output, etc. For Google: temperature, top_k, max_output_tokens, json_output, etc. type: object title: LLMLabeling description: |- Configuration for LLM-based cluster labeling. Supports multiple LLM providers with comprehensive model selection: - OpenAI: GPT-4o, GPT-4o-mini, GPT-4.1, O3-mini (best for quality) - Google: Gemini 2.5 Flash, Gemini 1.5 Flash (best for speed and cost) - Anthropic: Claude 3.5 Sonnet, Claude 3.5 Haiku (best for reasoning) All models are defined as enums and validated at API level. examples: - description: Text-only labeling with multiple fields enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: title path: title source_type: payload - input_key: description path: description source_type: payload - input_key: text path: text source_type: payload model_name: gpt-4o-mini-2024-07-18 provider: openai - description: Multimodal labeling with images (Gemini) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: headline source_type: payload - input_key: image_url path: thumbnail_url source_type: payload model_name: gemini-2.5-flash-lite provider: google - description: Multimodal labeling with video (Gemini) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: description source_type: payload - input_key: video_url path: video_url source_type: payload model_name: gemini-2.5-flash-lite provider: google - description: OpenAI GPT-4o (highest quality, text-only) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: gpt-4o-2024-08-06 provider: openai - description: Anthropic Claude 3.5 Sonnet (best reasoning) enabled: true include_keywords: true include_summary: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: claude-sonnet-4-5-20250929 provider: anthropic - description: 'Minimal configuration (uses defaults: text field from payload)' enabled: true - custom_prompt: >- You are a medical document classifier. Analyze the following patient record clusters and generate HIPAA-compliant category labels (2-3 words) that describe the medical condition or treatment type. DO NOT include patient names or identifiers. Return JSON array: [{"cluster_id": "cl_0", "label": "...", "keywords": [...]}] description: Custom prompt for domain-specific labeling enabled: true labeling_inputs: input_mappings: - input_key: text path: text source_type: payload model_name: gpt-4o-mini-2024-07-18 provider: openai SourceEnrichmentConfig: properties: field_mappings: items: $ref: '#/components/schemas/EnrichmentFieldMapping' type: array title: Field Mappings description: >- List of field mappings from cluster results to document fields. Default includes cluster_id and cluster_label. Can include: distance_to_centroid, member_count, keywords, visualization coords (x, y, z), etc. type: object title: SourceEnrichmentConfig description: >- Configuration for enriching source collection documents with cluster assignments. When enrich_source_collection=True, cluster assignments are written back to the original source documents, similar to taxonomy enrichment. Uses flexible field mapping pattern to support any cluster result fields. examples: - field_mappings: - source_field: cluster_id target_field: category_id - source_field: cluster_label target_field: category_name - source_field: distance_to_centroid target_field: category_confidence - field_mappings: - source_field: cluster_id target_field: segment - source_field: cluster_label target_field: segment_label - source_field: x target_field: viz_x - source_field: 'y' target_field: viz_y ClusterStats: properties: num_clusters: type: integer title: Num Clusters noise_points: anyOf: - type: integer - type: 'null' title: Noise Points description: Number of noise points (for DBSCAN, etc.) silhouette_score: anyOf: - type: number - type: 'null' title: Silhouette Score description: Silhouette score (-1 to 1, higher is better) extra: additionalProperties: true type: object title: Extra type: object required: - num_clusters title: ClusterStats description: Basic clustering quality metrics. TaskStatusEnum: type: string enum: - PENDING - QUEUED - IN_PROGRESS - PROCESSING - COMPLETED - COMPLETED_WITH_ERRORS - FAILED - CANCELED - INTERRUPTED - UNKNOWN - SKIPPED - DRAFT - ACTIVE - ARCHIVED - SUSPENDED title: TaskStatusEnum description: |- Enumeration of task statuses for tracking asynchronous operations. Task statuses indicate the current state of asynchronous operations like batch processing, object ingestion, clustering, and taxonomy execution. Status Categories: Operation Statuses: Track progress of async operations Lifecycle Statuses: Track entity state (buckets, collections, namespaces) Values: PENDING: Task is queued but has not started processing yet IN_PROGRESS: Task is currently being executed PROCESSING: Task is actively processing data (similar to IN_PROGRESS) COMPLETED: Task finished successfully with no errors COMPLETED_WITH_ERRORS: Task finished but some items failed (partial success) FAILED: Task encountered an error and could not complete CANCELED: Task was manually canceled by a user or system UNKNOWN: Task status could not be determined SKIPPED: Task was intentionally skipped DRAFT: Task is in draft state and not yet submitted ACTIVE: Entity is active and operational (for buckets, collections, etc.) ARCHIVED: Entity has been archived SUSPENDED: Entity has been temporarily suspended Terminal Statuses: COMPLETED, COMPLETED_WITH_ERRORS, FAILED, CANCELED are terminal statuses. Once a task reaches these states, it will not transition to another state. Partial Success Handling: COMPLETED_WITH_ERRORS indicates that the operation completed but some documents/items failed. The task result includes: - List of successful items - List of failed items with error details - Success rate percentage This allows clients to handle partial success scenarios appropriately. Polling Guidance: - Poll tasks in PENDING, QUEUED, IN_PROGRESS, or PROCESSING states - Stop polling when task reaches COMPLETED, COMPLETED_WITH_ERRORS, FAILED, or CANCELED - Use exponential backoff (1s ā†’ 30s) when polling ErrorDetail: properties: message: type: string title: Message description: Human-readable error message type: type: string title: Type description: Stable error type identifier (machine-readable) code: anyOf: - type: string - type: 'null' title: Code description: >- Fine-grained error code for programmatic handling (e.g., namespace_name_taken, feature_extractor_not_found). Present only when consumers may need to branch on a specific error condition. details: anyOf: - additionalProperties: true type: object - type: 'null' title: Details description: >- Optional structured details to help debugging (validation errors, IDs, etc.) type: object required: - message - type title: ErrorDetail description: Error detail model. ValidationError: properties: loc: items: anyOf: - type: string - type: integer type: array title: Location msg: type: string title: Message type: type: string title: Error Type type: object required: - loc - msg - type title: ValidationError LLMLabelingInput-Input: properties: input_mappings: items: $ref: '#/components/schemas/InputMapping' type: array minItems: 1 title: Input Mappings description: >- Flexible input mappings for constructing LLM context. Supports multimodal inputs (text, image_url, video_url, audio_url). Each mapping specifies how to extract data from document payloads. At least one input mapping is required. type: object required: - input_mappings title: LLMLabelingInput description: |- Input configuration for LLM-based cluster labeling. Supports flexible input mappings similar to retrievers and buckets, allowing multimodal inputs (text, images, videos, audio) for providers like Gemini that support native multimodal understanding. Examples: # Text-only labeling: LLMLabelingInput(input_mappings=[ InputMapping(input_key="headline", source_type="payload", path="headline"), InputMapping(input_key="description", source_type="payload", path="description") ]) # Multimodal labeling with images: LLMLabelingInput(input_mappings=[ InputMapping(input_key="text", source_type="payload", path="headline"), InputMapping(input_key="image_url", source_type="payload", path="thumbnail_url") ]) # Multimodal with video (for Gemini): LLMLabelingInput(input_mappings=[ InputMapping(input_key="text", source_type="payload", path="description"), InputMapping(input_key="video_url", source_type="payload", path="video_url") ]) LLMProvider: type: string enum: - openai - google - anthropic title: LLMProvider description: >- Supported LLM providers for content generation. Each provider has different strengths, pricing, and multimodal capabilities. Choose based on your use case, performance requirements, and budget. Values: OPENAI: OpenAI GPT models (GPT-4o, GPT-4.1, O3-mini) - Best for: General purpose, vision tasks, structured outputs - Multimodal: Text, images - Performance: Fast (100-500ms), reliable - Cost: Moderate to high ($0.15-$10 per 1M tokens) - Use when: Need high-quality generation with vision support GOOGLE: Google Gemini models (Gemini 3.1 Flash Lite, Gemini 2.5 Pro) - Best for: Fast generation, video understanding, cost-efficiency - Multimodal: Text, images, video, audio, PDFs - Performance: Very fast (50-200ms) - Cost: Low to moderate ($0.075-$0.40 per 1M tokens) - Use when: Need video/audio/PDF support or cost-efficiency ANTHROPIC: Anthropic Claude models (Claude 3.5 Sonnet, Claude 3.5 Haiku) - Best for: Long context, complex reasoning, safety - Multimodal: Text, images - Performance: Moderate (200-800ms) - Cost: Moderate to high ($0.25-$15 per 1M tokens) - Use when: Need long context or complex reasoning Examples: - Use OPENAI for production with structured JSON outputs - Use GOOGLE for video summarization and cost-sensitive workloads - Use ANTHROPIC for complex reasoning with long documents OpenAIModel: type: string enum: - gpt-4o-2024-08-06 - gpt-4o-mini-2024-07-18 - gpt-4.1-2025-04-14 - gpt-4.1-mini-2025-04-14 - o3-mini-2025-01-31 title: OpenAIModel description: |- OpenAI model identifiers for LLM generation. Models listed in order of capability and cost (highest to lowest). All models support vision (images) except O3-mini. Values: GPT_4O: Latest GPT-4 Omni model (2024-08-06) - Use for: Production, highest quality generation - Context: 128K tokens - Vision: Yes - Cost: $2.50/1M input, $10/1M output - Performance: 200-500ms per request - When to use: Need best quality, willing to pay premium GPT_41: GPT-4.1 (2025-04-14) - Use for: Future-proofed pipelines - Context: 128K tokens - Vision: Yes - Cost: TBD (expected similar to GPT-4o) - When to use: Want latest model features GPT_4O_MINI: Smaller, faster GPT-4 Omni (2024-07-18) - Use for: High-volume, cost-sensitive workloads - Context: 128K tokens - Vision: Yes - Cost: $0.15/1M input, $0.60/1M output - Performance: 100-200ms per request - When to use: Good balance of quality and cost GPT_41_MINI: Smaller GPT-4.1 (2025-04-14) - Use for: Future cost-optimized pipelines - Context: 128K tokens - Vision: Yes - Cost: TBD (expected similar to GPT-4o-mini) - When to use: Want latest features at lower cost O3_MINI: Reasoning-optimized model (2025-01-31) - Use for: Complex reasoning, math, code - Context: 200K tokens - Vision: No - Cost: TBD - When to use: Need advanced reasoning capabilities Examples: - Use GPT_4O for caption generation with images (best quality) - Use GPT_4O_MINI for high-volume video scene summarization (cost-effective) - Use O3_MINI for complex entity extraction requiring reasoning GoogleModel: type: string enum: - gemini-2.5-flash-lite - gemini-2.5-flash - gemini-2.5-pro - gemini-3.1-flash-lite title: GoogleModel description: >- Google Gemini model identifiers for LLM generation. Gemini models excel at multimodal understanding with best-in-class video support. All models support text, images, video, audio, and PDFs. Values: GEMINI_2_5_FLASH_LITE: Gemini 2.5 Flash Lite model (recommended, stable GA) - Use for: Fastest generation, cost-effective multimodal - Context: 1M tokens - Multimodal: Text, images, video, audio, PDFs - When to use: Default choice for all Gemini use cases GEMINI_2_5_PRO: Gemini 2.5 Pro model - Use for: Higher quality reasoning, complex tasks - Context: 1M tokens GEMINI_2_5_FLASH: Gemini 2.5 Flash model - Kept for backward compatibility. GEMINI_3_1_FLASH_LITE: Alias for gemini-2.5-flash-lite (backwards compat) - Note: gemini-3.1-flash-lite does NOT exist in Google's API. This value is mapped to gemini-2.5-flash-lite at runtime. AnthropicModel: type: string enum: - claude-sonnet-4-5-20250929 - claude-haiku-4-5-20251001 - claude-3-5-sonnet-20241022 - claude-3-5-haiku-20241022 title: AnthropicModel description: |- Anthropic Claude model identifiers for LLM generation. Claude models excel at long context, complex reasoning, and safety. All models support text and images. Values: CLAUDE_3_5_SONNET: Most capable Claude model - Use for: Complex reasoning, long documents, safety-critical - Context: 200K tokens - Vision: Yes - Cost: $3/1M input, $15/1M output - Performance: 300-800ms per request - When to use: Need best reasoning, safety, or long context CLAUDE_3_5_HAIKU: Fast, cost-effective Claude model - Use for: High-volume, quick summaries - Context: 200K tokens - Vision: Yes - Cost: $0.25/1M input, $1.25/1M output - Performance: 100-300ms per request - When to use: Good balance of quality and cost Examples: - Use CLAUDE_3_5_SONNET for complex entity extraction from contracts (best reasoning) - Use CLAUDE_3_5_HAIKU for high-volume content moderation (cost-effective) ClusteringAlgorithm: type: string enum: - kmeans - dbscan - hdbscan - agglomerative - spectral - gaussian_mixture - mean_shift - optics - leiden - attribute_based - auto title: ClusteringAlgorithm description: |- Supported clustering algorithms. Two types of clustering are available: 1. Vector-based: Clusters documents by embedding similarity 2. Attribute-based: Clusters documents by metadata attributes Vector-based algorithms (require feature_vector): - kmeans: Partitions data into K clusters by minimizing within-cluster variance - dbscan: Density-based clustering, finds clusters of arbitrary shape - hdbscan: Hierarchical DBSCAN, auto-determines number of clusters - agglomerative: Hierarchical clustering using linkage criteria - spectral: Uses graph theory to find clusters - gaussian_mixture: Probabilistic model assuming Gaussian distributions - mean_shift: Finds clusters by locating density maxima - optics: Ordering points to identify clustering structure - leiden: Community detection on a kNN graph (fast at scale, auto cluster count via resolution; skips the UMAP step that dominates KMeans) Attribute-based algorithm (requires attribute_config): - attribute_based: Groups documents by metadata attributes (e.g., category, brand) KMeansParams: properties: n_clusters: type: integer maximum: 1000 minimum: 2 title: N Clusters description: Number of clusters to form default: 8 max_iter: type: integer maximum: 10000 minimum: 1 title: Max Iter description: Maximum number of iterations default: 300 random_state: anyOf: - type: integer - type: 'null' title: Random State description: Random seed for reproducibility default: 42 n_init: type: integer minimum: 1 title: N Init description: Number of times k-means will run with different centroid seeds default: 10 tol: type: number exclusiveMinimum: 0 title: Tol description: Tolerance for convergence default: 0.0001 init: type: string title: Init description: Method for initialization ('k-means++' or 'random') default: k-means++ verbose: type: integer minimum: 0 title: Verbose description: Verbosity mode default: 0 copy_x: type: boolean title: Copy X description: If True, the original data is not modified default: true algorithm: type: string title: Algorithm description: K-means algorithm to use ('lloyd', 'elkan', or 'auto') default: lloyd type: object title: KMeansParams description: Parameters for K-Means clustering algorithm. DBSCANParams: properties: eps: type: number exclusiveMinimum: 0 title: Eps description: >- Maximum distance between two samples for one to be considered in the neighborhood of the other default: 0.5 min_samples: type: integer minimum: 1 title: Min Samples description: >- Number of samples in a neighborhood for a point to be considered a core point default: 5 metric: type: string title: Metric description: Metric to use for distance computation default: euclidean metric_params: additionalProperties: true type: object title: Metric Params description: Additional keyword arguments for the metric function algorithm: type: string title: Algorithm description: >- Algorithm to compute pointwise distances and find nearest neighbors ('auto', 'ball_tree', 'kd_tree', 'brute') default: auto leaf_size: type: integer minimum: 1 title: Leaf Size description: Leaf size passed to BallTree or KDTree default: 30 p: type: number exclusiveMinimum: 0 title: P description: >- The power of the Minkowski metric to be used to calculate distance between points default: 2 n_jobs: type: integer title: N Jobs description: The number of parallel jobs to run (-1 means using all processors) default: 1 type: object title: DBSCANParams description: Parameters for DBSCAN clustering algorithm. HDBSCANParams: properties: min_cluster_size: type: integer minimum: 2 title: Min Cluster Size description: Minimum number of samples in a cluster default: 5 min_samples: anyOf: - type: integer minimum: 1 - type: 'null' title: Min Samples description: >- Number of samples in a neighborhood for a point to be considered a core point. Defaults to min_cluster_size if None cluster_selection_epsilon: type: number minimum: 0 title: Cluster Selection Epsilon description: >- A distance threshold for cluster selection. Clusters below this value will be merged default: 0 max_cluster_size: anyOf: - type: integer minimum: 1 - type: 'null' title: Max Cluster Size description: >- Maximum number of samples in a cluster. Clusters above this size will be split metric: type: string title: Metric description: Metric to use for distance computation default: euclidean alpha: type: number exclusiveMinimum: 0 title: Alpha description: A distance scaling parameter default: 1 cluster_selection_method: type: string title: Cluster Selection Method description: Method to select clusters from the condensed tree ('eom' or 'leaf') default: eom allow_single_cluster: type: boolean title: Allow Single Cluster description: Allow HDBSCAN to find only a single cluster default: false prediction_data: type: boolean title: Prediction Data description: Whether to generate extra data for predicting cluster membership default: false match_reference_implementation: type: boolean title: Match Reference Implementation description: Whether to match the reference implementation exactly default: false type: object title: HDBSCANParams description: Parameters for HDBSCAN clustering algorithm. AgglomerativeParams: properties: n_clusters: anyOf: - type: integer minimum: 2 - type: 'null' title: N Clusters description: >- Number of clusters to find. Can be None if distance_threshold is not None default: 2 affinity: type: string title: Affinity description: >- Metric used to compute linkage. Can be 'euclidean', 'l1', 'l2', 'manhattan', 'cosine', or 'precomputed' default: euclidean memory: anyOf: - type: string - type: 'null' title: Memory description: Path to the caching directory connectivity: anyOf: - {} - type: 'null' title: Connectivity description: Connectivity matrix. Defines which samples are neighbors compute_full_tree: type: string title: Compute Full Tree description: Whether to compute the full tree ('auto', True, or False) default: auto linkage: type: string title: Linkage description: Linkage criterion ('ward', 'complete', 'average', 'single') default: ward distance_threshold: anyOf: - type: number exclusiveMinimum: 0 - type: 'null' title: Distance Threshold description: >- The linkage distance threshold above which clusters will not be merged compute_distances: type: boolean title: Compute Distances description: Whether to compute distances between clusters default: false type: object title: AgglomerativeParams description: Parameters for Agglomerative clustering algorithm. SpectralParams: properties: n_clusters: type: integer minimum: 2 title: N Clusters description: Number of clusters to form default: 8 eigen_solver: anyOf: - type: string - type: 'null' title: Eigen Solver description: >- The eigenvalue decomposition strategy ('arpack', 'lobpcg', 'amg', or None) n_components: anyOf: - type: integer minimum: 1 - type: 'null' title: N Components description: Number of eigenvectors to use for spectral embedding random_state: anyOf: - type: integer - type: 'null' title: Random State description: Random seed for reproducibility default: 42 n_init: type: integer minimum: 1 title: N Init description: Number of times k-means will run with different centroid seeds default: 10 gamma: type: number exclusiveMinimum: 0 title: Gamma description: >- Kernel coefficient for rbf, poly, sigmoid, laplacian and chi2 kernels default: 1 affinity: type: string title: Affinity description: >- How to construct the affinity matrix ('nearest_neighbors', 'rbf', 'precomputed', 'precomputed_nearest_neighbors') default: rbf n_neighbors: type: integer minimum: 1 title: N Neighbors description: >- Number of neighbors to use when constructing the affinity matrix using nearest neighbors default: 10 eigen_tol: type: number minimum: 0 title: Eigen Tol description: Stopping criterion for eigendecomposition default: 0 assign_labels: type: string title: Assign Labels description: >- Strategy to assign labels in the embedding space ('kmeans' or 'discretize') default: kmeans degree: type: number title: Degree description: Degree of the polynomial kernel. Ignored by other kernels default: 3 coef0: type: number title: Coef0 description: Zero coefficient for polynomial and sigmoid kernels default: 1 kernel_params: anyOf: - additionalProperties: true type: object - type: 'null' title: Kernel Params description: Parameters for the kernel function n_jobs: type: integer title: N Jobs description: Number of parallel jobs to run (-1 means using all processors) default: 1 verbose: type: boolean title: Verbose description: Verbosity mode default: false type: object title: SpectralParams description: Parameters for Spectral clustering algorithm. GaussianMixtureParams: properties: n_components: type: integer minimum: 1 title: N Components description: Number of mixture components default: 1 covariance_type: type: string title: Covariance Type description: Type of covariance parameters ('full', 'tied', 'diag', 'spherical') default: full tol: type: number exclusiveMinimum: 0 title: Tol description: Convergence threshold default: 0.001 reg_covar: type: number minimum: 0 title: Reg Covar description: Regularization added to the diagonal of covariance default: 0.000001 max_iter: type: integer minimum: 1 title: Max Iter description: Maximum number of EM iterations default: 100 n_init: type: integer minimum: 1 title: N Init description: Number of initializations to perform default: 1 init_params: type: string title: Init Params description: >- Method used to initialize weights, means and covariances ('kmeans' or 'random') default: kmeans weights_init: anyOf: - items: {} type: array - type: 'null' title: Weights Init description: Initial weights means_init: anyOf: - items: {} type: array - type: 'null' title: Means Init description: Initial means precisions_init: anyOf: - items: {} type: array - type: 'null' title: Precisions Init description: Initial precisions random_state: anyOf: - type: integer - type: 'null' title: Random State description: Random seed for reproducibility default: 42 warm_start: type: boolean title: Warm Start description: If True, use the solution of the last fit as initialization default: false verbose: type: integer minimum: 0 title: Verbose description: Enable verbose output default: 0 verbose_interval: type: integer minimum: 1 title: Verbose Interval description: Number of iterations between each verbose message default: 10 type: object title: GaussianMixtureParams description: Parameters for Gaussian Mixture Model clustering. MeanShiftParams: properties: bandwidth: anyOf: - type: number exclusiveMinimum: 0 - type: 'null' title: Bandwidth description: >- Bandwidth used in the RBF kernel. If None, estimated using sklearn.cluster.estimate_bandwidth seeds: anyOf: - items: items: type: number type: array type: array - type: 'null' title: Seeds description: >- Seeds used to initialize kernels. If None, all points are used as seeds bin_seeding: type: boolean title: Bin Seeding description: >- If true, initial kernel locations are discretized into a grid to speed up algorithm default: false min_bin_freq: type: integer minimum: 1 title: Min Bin Freq description: Minimum number of seeds within a bin for the bin to be considered default: 1 cluster_all: type: boolean title: Cluster All description: >- If true, all points are clustered, even orphans. If false, orphans are given label -1 default: true n_jobs: type: integer title: N Jobs description: Number of parallel jobs to run (-1 means using all processors) default: 1 max_iter: type: integer minimum: 1 title: Max Iter description: >- Maximum number of iterations per seed point before the algorithm stops default: 300 type: object title: MeanShiftParams description: Parameters for Mean Shift clustering algorithm. OPTICSParams: properties: min_samples: type: integer minimum: 2 title: Min Samples description: >- Number of samples in a neighborhood for a point to be considered a core point default: 5 max_eps: anyOf: - type: number exclusiveMinimum: 0 - type: 'null' title: Max Eps description: >- Maximum distance between two samples. Default (None) means no maximum distance metric: type: string title: Metric description: Metric to use for distance computation default: minkowski p: type: number exclusiveMinimum: 0 title: P description: Parameter for the Minkowski metric default: 2 metric_params: anyOf: - additionalProperties: true type: object - type: 'null' title: Metric Params description: Additional keyword arguments for the metric function cluster_method: type: string title: Cluster Method description: Method to extract clusters ('xi' or 'dbscan') default: xi eps: anyOf: - type: number exclusiveMinimum: 0 - type: 'null' title: Eps description: Maximum distance for DBSCAN cluster extraction method xi: type: number maximum: 1 minimum: 0 title: Xi description: >- Minimum steepness on the reachability plot for cluster boundary (xi method) default: 0.05 predecessor_correction: type: boolean title: Predecessor Correction description: Correct clusters based on predecessors (xi method) default: true min_cluster_size: anyOf: - type: number - type: 'null' title: Min Cluster Size description: Minimum number of samples in a cluster. Can be a fraction if < 1.0 algorithm: type: string title: Algorithm description: >- Algorithm to compute pointwise distances ('auto', 'ball_tree', 'kd_tree', 'brute') default: auto leaf_size: type: integer minimum: 1 title: Leaf Size description: Leaf size passed to BallTree or KDTree default: 30 n_jobs: type: integer title: N Jobs description: Number of parallel jobs to run (-1 means using all processors) default: 1 type: object title: OPTICSParams description: Parameters for OPTICS clustering algorithm. LeidenParams: properties: resolution: type: number maximum: 100 exclusiveMinimum: 0 title: Resolution description: >- Resolution for the Leiden objective. Higher values yield more, smaller communities; lower values yield fewer, larger ones. Leiden discovers the cluster count from the graph ā€” there is no fixed n_clusters. default: 1 n_neighbors: type: integer maximum: 200 minimum: 2 title: N Neighbors description: >- Number of nearest neighbours per node when building the kNN graph. Larger values give a denser graph (smoother communities) at higher build cost. default: 30 metric: type: string title: Metric description: Distance metric for kNN graph construction (cosine or euclidean). default: cosine min_cluster_size: type: integer minimum: 0 title: Min Cluster Size description: >- Communities smaller than this are relabelled as noise (cluster_id -1). 0 keeps every community. default: 0 objective_function: type: string enum: - RBConfiguration - modularity - CPM title: Objective Function description: >- Leiden quality function. RBConfiguration and CPM honour the resolution parameter; modularity ignores it. default: RBConfiguration n_iterations: type: integer title: N Iterations description: >- Leiden optimisation passes. -1 runs until no further improvement; a small positive value (e.g. 2) is faster and usually sufficient. default: 2 random_state: anyOf: - type: integer - type: 'null' title: Random State description: Random seed for reproducible partitions. default: 42 type: object title: LeidenParams description: Parameters for Leiden graph community-detection clustering. WeightLearningConfig: properties: method: type: string enum: - grid_search - bayesian title: Method description: >- Weight learning method: - bayesian: Gaussian process optimization (recommended, scales to 5+ features) - grid_search: Exhaustive search (limited to 2-3 features, simpler but slower) default: bayesian examples: - bayesian - grid_search max_iterations: type: integer maximum: 100 minimum: 5 title: Max Iterations description: >- Maximum optimization iterations: - grid_search: Number of values to try per feature (total: max_iterations^n_features) - bayesian: Number of weight combinations to evaluate Recommended: 20 for bayesian, 5 for grid_search default: 20 examples: - 20 - 50 metric: type: string enum: - silhouette - davies_bouldin - calinski_harabasz title: Metric description: >- Clustering quality metric to optimize: - silhouette: Measures how similar points are to their cluster vs other clusters (range: [-1, 1], higher is better) - davies_bouldin: Ratio of within-cluster to between-cluster distances (range: [0, āˆž], lower is better) - calinski_harabasz: Ratio of between-cluster to within-cluster variance (range: [0, āˆž], higher is better) Recommended: silhouette (most general-purpose) default: silhouette examples: - silhouette - davies_bouldin - calinski_harabasz sample_size: anyOf: - type: integer minimum: 100 - type: 'null' title: Sample Size description: >- Optional: Learn weights on a random sample (speeds up large datasets). If provided and dataset has more documents, weights are learned on sample_size random documents, then applied to full dataset. Recommended: 5000 for datasets >10k documents examples: - 5000 - 10000 random_state: type: integer title: Random State description: Random seed for reproducibility of weight learning default: 42 examples: - 42 type: object title: WeightLearningConfig description: >- Configuration for automatic feature weight learning in multi-feature clustering. When multi_feature_strategy='weighted' and feature_weights is not provided, this configuration controls how optimal weights are automatically learned. The system tries different weight combinations and picks the one that produces the best clustering quality (measured by silhouette score, etc.). Examples: Bayesian optimization (recommended): { "method": "bayesian", "max_iterations": 20, "metric": "silhouette", "sample_size": 5000 } Grid search (exhaustive, limited to 2-3 features): { "method": "grid_search", "max_iterations": 5, "metric": "silhouette" } examples: - max_iterations: 20 method: bayesian metric: silhouette random_state: 42 sample_size: 5000 - max_iterations: 5 method: grid_search metric: silhouette random_state: 42 TSNEParams: properties: method: type: string const: tsne title: Method default: tsne n_components: type: integer maximum: 256 minimum: 2 title: N Components default: 2 random_state: type: integer title: Random State default: 42 perplexity: type: number exclusiveMinimum: 0 title: Perplexity default: 30 learning_rate: type: number exclusiveMinimum: 0 title: Learning Rate default: 200 type: object title: TSNEParams UMAPParams: properties: method: type: string const: umap title: Method default: umap n_components: type: integer maximum: 256 minimum: 2 title: N Components default: 50 random_state: type: integer title: Random State default: 42 n_neighbors: type: integer minimum: 2 title: N Neighbors default: 30 min_dist: type: number maximum: 1 minimum: 0 title: Min Dist default: 0 metric: type: string title: Metric description: >- Distance metric for UMAP. 'cosine' is best for normalized embeddings. default: cosine type: object title: UMAPParams WhiteningParams: properties: method: type: string const: whitening title: Method default: whitening regularization: type: number minimum: 0 title: Regularization description: Eigenvalue floor to prevent division by near-zero values. default: 0.00001 type: object title: WhiteningParams NoReduction: properties: method: type: string const: none title: Method default: none type: object title: NoReduction FilterCondition: properties: field: type: string title: Field description: Field name to filter on operator: $ref: '#/components/schemas/FilterOperator' description: Comparison operator default: eq value: anyOf: - $ref: '#/components/schemas/DynamicValue' - {} title: Value description: Value to compare against type: object required: - field - value title: FilterCondition description: |- Represents a single filter condition. Attributes: field: The field to filter on operator: The comparison operator value: The value to compare against LLMLabelingInput-Output: properties: input_mappings: items: $ref: '#/components/schemas/InputMapping' type: array minItems: 1 title: Input Mappings description: >- Flexible input mappings for constructing LLM context. Supports multimodal inputs (text, image_url, video_url, audio_url). Each mapping specifies how to extract data from document payloads. At least one input mapping is required. type: object required: - input_mappings title: LLMLabelingInput description: |- Input configuration for LLM-based cluster labeling. Supports flexible input mappings similar to retrievers and buckets, allowing multimodal inputs (text, images, videos, audio) for providers like Gemini that support native multimodal understanding. Examples: # Text-only labeling: LLMLabelingInput(input_mappings=[ InputMapping(input_key="headline", source_type="payload", path="headline"), InputMapping(input_key="description", source_type="payload", path="description") ]) # Multimodal labeling with images: LLMLabelingInput(input_mappings=[ InputMapping(input_key="text", source_type="payload", path="headline"), InputMapping(input_key="image_url", source_type="payload", path="thumbnail_url") ]) # Multimodal with video (for Gemini): LLMLabelingInput(input_mappings=[ InputMapping(input_key="text", source_type="payload", path="description"), InputMapping(input_key="video_url", source_type="payload", path="video_url") ]) EnrichmentFieldMapping: properties: source_field: type: string title: Source Field description: >- Field from cluster results to include. Available fields: cluster_id, cluster_label, distance_to_centroid, member_count, keywords, x, y, z (visualization coords), metadata.* target_field: type: string title: Target Field description: >- Target field name in enriched document. Example: 'category_id' for cluster_id, 'product_category' for cluster_label type: object required: - source_field - target_field title: EnrichmentFieldMapping description: |- Maps a cluster result field to a document enrichment field. Similar to InputMapping pattern used throughout Mixpeek. InputMapping: properties: input_key: type: string title: Input Key description: Key used in the constructed inputs payload. source_type: anyOf: - $ref: '#/components/schemas/InputSourceType' - type: 'null' description: Source of the value (payload, literal, vector). path: anyOf: - type: string - type: 'null' title: Path description: >- Dot-notation path inside payload/vector when source_type is PAYLOAD or VECTOR. override: anyOf: - {} - type: 'null' title: Override description: Static value used when source_type is LITERAL. Overrides any path. type: object required: - input_key title: InputMapping description: |- Declarative mapping for building inputs from various sources. - input_key: The key used in the constructed inputs payload - source_type: Where to fetch the value (payload, literal, vector) - path: Dot-notation path when source_type is PAYLOAD or VECTOR - override: Static value when source_type is LITERAL examples: - input_key: query_text path: content.title source_type: payload - input_key: lang override: en source_type: literal - input_key: image_vector path: features.clip_vit_l_14 source_type: vector FilterOperator: type: string enum: - eq - ne - gt - lt - gte - lte - in - nin - contains - starts_with - ends_with - regex - exists - is_null - text - phrase title: FilterOperator description: Supported filter operators across database implementations. DynamicValue: properties: type: type: string const: dynamic title: Type default: dynamic field: type: string title: Field description: >- The dot-notation path to the value in the runtime query request, e.g., 'inputs.user_id' examples: - inputs.query_text - filters.AND[0].value type: object required: - field title: DynamicValue description: A value that should be dynamically resolved from the query request. InputSourceType: type: string enum: - payload - literal - vector - blob title: InputSourceType description: Where the value for an input should be retrieved from. ````