{"passport":{"unfragile":{"@version":"1.0","version":"2026-05","artifact":{"id":"nomic-embed","slug":"nomic-embed","name":"Nomic Embed","type":"repo","url":"https://github.com/nomic-ai/nomic","page_url":"https://unfragile.ai/nomic-embed","categories":["model-training","rag-knowledge","testing-quality"],"tags":[],"pricing":{"model":"free","free":true,"starting_price":null},"status":"active","verified":false},"capabilities":[{"id":"nomic-embed__cap_0","uri":"capability://data.processing.analysis.matryoshka.based.multi.scale.text.embedding.generation","name":"matryoshka-based multi-scale text embedding generation","description":"Generates dense vector embeddings for text using Matryoshka representation learning, which produces nested embeddings at multiple dimensionalities (e.g., 768, 512, 256, 128 dimensions) from a single forward pass. This allows downstream consumers to trade off between embedding quality and computational cost by selecting the appropriate dimensionality without recomputing. The architecture uses transformer-based models trained with contrastive objectives to preserve semantic relationships across all scales.","intents":["Generate embeddings for large text corpora while maintaining flexibility to adjust vector dimensionality based on memory/latency constraints","Build RAG systems that can dynamically choose embedding dimensions for different query types or retrieval stages","Reduce storage and inference costs by using lower-dimensional embeddings for less critical retrieval operations"],"best_for":["Teams building production RAG systems with strict latency or memory budgets","Researchers exploring multi-scale semantic representations","Organizations processing massive text datasets where embedding storage is a bottleneck"],"limitations":["Matryoshka training adds complexity to fine-tuning workflows compared to fixed-dimension models","Quality degradation increases at lower dimensionalities; 128-dim embeddings may lose semantic precision for nuanced queries","No built-in adaptive selection mechanism — applications must implement their own logic to choose dimensionality per query"],"requires":["Python 3.8+","PyTorch 1.9+ for model inference","GPU recommended for batch embedding generation (CPU inference ~10-50x slower)"],"input_types":["text (strings, documents, paragraphs)","batched text arrays"],"output_types":["dense float32 vectors (variable dimensionality: 768, 512, 256, 128, etc.)","structured embedding metadata (dimensionality, model version)"],"categories":["data-processing-analysis","embedding-generation"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_1","uri":"capability://data.processing.analysis.multimodal.embedding.generation.for.text.and.images","name":"multimodal embedding generation for text and images","description":"Generates joint embeddings for both text and image inputs in a shared vector space, enabling cross-modal semantic search and similarity matching. The implementation uses a dual-encoder architecture where text and image encoders are trained with contrastive objectives to align their representations. Supports both pre-computed image embeddings and raw image inputs, with automatic image preprocessing and encoding.","intents":["Build multimodal search systems where users can query with text to find similar images or vice versa","Create unified embeddings for datasets containing both text descriptions and visual content","Implement content-based recommendation systems that leverage both textual and visual features"],"best_for":["E-commerce and product discovery teams building visual search","Content platforms (news, social media) needing cross-modal search","Researchers working with multimodal datasets requiring unified representations"],"limitations":["Image encoding adds 50-200ms per image depending on resolution and hardware","Alignment quality depends on training data diversity; performance may degrade on domain-specific images not well-represented in training set","No built-in support for video or 3D data; only static images","Cross-modal retrieval performance typically 5-15% lower than single-modality search due to alignment trade-offs"],"requires":["Python 3.8+","PyTorch 1.9+","PIL/Pillow for image preprocessing","GPU strongly recommended (CPU image encoding ~20-50x slower)"],"input_types":["text (strings, documents)","images (PIL Image objects, file paths, or raw image bytes)","pre-computed image embeddings (numpy arrays)"],"output_types":["dense float32 vectors (shared embedding space, typically 768-dim)","image metadata (resolution, preprocessing details)"],"categories":["data-processing-analysis","image-visual"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_10","uri":"capability://tool.use.integration.shareable.interactive.map.urls.and.collaborative.exploration","name":"shareable interactive map urls and collaborative exploration","description":"Generates shareable URLs for Atlas maps that allow non-technical users to explore datasets interactively without installing software. The implementation creates web-based visualizations hosted on the Atlas platform with support for filtering, searching, and zooming. Maps can be shared with specific permissions (view-only, edit, etc.) and support collaborative annotations.","intents":["Share dataset explorations with stakeholders who don't have technical setup","Enable collaborative exploration and annotation of datasets across teams","Create public or private dashboards for dataset discovery and understanding"],"best_for":["Product and business teams exploring data without technical setup","Research teams sharing findings with collaborators","Organizations creating public data explorations for transparency"],"limitations":["Web UI performance degrades with >100k data points; may require data sampling for large datasets","Sharing permissions are managed through Nomic platform; no integration with enterprise SSO","Collaborative annotations are not version-controlled; conflicts possible with concurrent edits","Maps are hosted on Nomic servers; no option for self-hosted visualizations"],"requires":["Python 3.8+ (for creating maps)","Web browser for viewing (Chrome, Firefox, Safari)","Nomic account for sharing and permissions"],"input_types":["AtlasProjection object (created via atlas.map_data())","sharing permissions (view, edit, etc.)"],"output_types":["shareable URL","embedded visualization code","permission tokens"],"categories":["tool-use-integration","memory-knowledge"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_11","uri":"capability://automation.workflow.aws.sagemaker.and.pytorch.lightning.integration.for.distributed.training","name":"aws sagemaker and pytorch lightning integration for distributed training","description":"Provides integration with AWS SageMaker for distributed model training and PyTorch Lightning for streamlined training workflows. The implementation includes pre-configured training scripts and configuration files that enable fine-tuning Nomic models on custom datasets at scale. Supports distributed training across multiple GPUs and nodes with automatic checkpointing and logging.","intents":["Fine-tune Nomic embedding models on proprietary datasets using managed AWS infrastructure","Scale training across multiple GPUs without managing infrastructure","Integrate model training into MLOps pipelines with automatic logging and checkpointing"],"best_for":["Teams with AWS infrastructure wanting to fine-tune models at scale","ML engineers building custom embedding models for domain-specific applications","Organizations with large proprietary datasets requiring model customization"],"limitations":["AWS SageMaker integration requires AWS account and knowledge of SageMaker APIs","Training costs scale with compute resources; can be expensive for large models and datasets","PyTorch Lightning requires familiarity with PyTorch and distributed training concepts","No built-in hyperparameter tuning; requires manual experimentation or external tools"],"requires":["Python 3.8+","PyTorch 1.9+","PyTorch Lightning 1.5+","AWS account with SageMaker access","Training dataset in S3 or local storage"],"input_types":["training dataset (text pairs, images)","training configuration (YAML/JSON)","pre-trained model checkpoint"],"output_types":["fine-tuned model checkpoint","training metrics and logs","model artifacts for deployment"],"categories":["automation-workflow","tool-use-integration"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_12","uri":"capability://tool.use.integration.gpt4all.integration.for.local.model.inference.and.fine.tuning","name":"gpt4all integration for local model inference and fine-tuning","description":"Integrates with GPT4All to enable local inference of embedding models without cloud dependencies or API keys. The implementation downloads quantized model weights and runs inference locally using optimized inference engines. Supports both CPU and GPU inference with automatic hardware detection.","intents":["Generate embeddings locally without sending data to external APIs for privacy-sensitive applications","Run embedding models on edge devices or offline environments","Reduce latency and costs by eliminating cloud API calls for embedding generation"],"best_for":["Organizations with strict data privacy requirements","Teams building offline-capable RAG systems","Edge computing scenarios where cloud connectivity is unavailable"],"limitations":["Local inference is 10-50x slower than GPU cloud inference depending on hardware","Quantized models may have slightly lower quality than full-precision models (typically <2% degradation)","Requires downloading model weights (500MB-2GB); initial setup takes time","CPU inference is practical only for small batch sizes (<10 documents)"],"requires":["Python 3.8+","GPT4All library (pip install gpt4all)","4GB+ RAM for model loading","GPU optional but recommended for reasonable performance"],"input_types":["text strings or lists","optional: batch size parameter"],"output_types":["embedding vectors","inference timing metadata"],"categories":["tool-use-integration","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_13","uri":"capability://tool.use.integration.gpt4all.integration.for.local.inference.without.api.keys","name":"gpt4all integration for local inference without api keys","description":"Integrates with GPT4All to enable local embedding inference without requiring API keys or cloud connectivity. The system provides compatibility layers that allow using Nomic embedding models through GPT4All's local inference engine, which runs models on CPU or GPU without external service calls. This enables offline embedding generation and privacy-preserving inference where data never leaves the user's machine.","intents":["Generate embeddings locally without sending data to cloud services or requiring API keys","Build privacy-preserving RAG systems where embeddings are computed on-device","Deploy embeddings in air-gapped environments without internet connectivity"],"best_for":["Organizations with strict data privacy requirements (healthcare, finance, government)","Teams building offline-first applications","Developers prototyping without API key setup"],"limitations":["Local inference is significantly slower than GPU-accelerated cloud inference (10-100x slower on CPU)","Requires sufficient local storage for model weights (typically 500MB-2GB per model)","CPU inference is memory-intensive; may not work on devices with <4GB RAM","No distributed inference across multiple machines; single-machine scaling only"],"requires":["Python 3.8+","GPT4All 1.0+","Local storage for model weights (500MB-2GB)","CPU or GPU for inference (GPU recommended for reasonable performance)"],"input_types":["text strings","lists of documents"],"output_types":["embedding vectors (numpy arrays)","embeddings in GPT4All format"],"categories":["tool-use-integration","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_2","uri":"capability://data.processing.analysis.full.training.data.transparency.and.reproducibility","name":"full training data transparency and reproducibility","description":"Provides complete documentation and access to training datasets, hyperparameters, and training procedures used to create embedding models. The architecture includes versioned dataset manifests, training configuration files, and reproducible training scripts that allow users to audit model provenance and retrain models with custom data. This enables transparency about potential biases and enables fine-tuning on domain-specific data.","intents":["Audit embedding models for bias and understand what data influenced their semantic representations","Fine-tune models on proprietary or domain-specific datasets using the same training methodology","Reproduce model training results for research or compliance purposes","Build trust in models by understanding their training data composition and potential limitations"],"best_for":["Regulated industries (finance, healthcare, legal) requiring model auditability","Research teams needing reproducible embedding models","Organizations with proprietary data wanting to fine-tune models without vendor lock-in"],"limitations":["Training data transparency may expose privacy concerns if datasets contain sensitive information","Reproducing training requires significant computational resources (GPU clusters, weeks of training time)","Fine-tuning requires expertise in machine learning and PyTorch; no low-code fine-tuning interface provided","Dataset documentation may be incomplete for edge cases or data quality issues"],"requires":["Python 3.8+","PyTorch 1.9+","Access to training dataset manifests (publicly available)","GPU cluster for retraining (optional but strongly recommended)"],"input_types":["training configuration files (YAML/JSON)","dataset manifests (CSV/JSON with data sources)","custom training data (text files, datasets)"],"output_types":["training logs and metrics","fine-tuned model checkpoints","reproducibility reports"],"categories":["data-processing-analysis","safety-moderation"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_3","uri":"capability://tool.use.integration.client.server.embedding.api.with.local.and.cloud.inference","name":"client-server embedding api with local and cloud inference","description":"Provides a Python client library that communicates with the Atlas platform backend to generate embeddings either locally (using downloaded models) or via cloud API endpoints. The architecture supports both synchronous and asynchronous embedding generation with batching, caching, and automatic fallback between local and cloud inference. Implements connection pooling and request queuing to optimize throughput for large-scale embedding jobs.","intents":["Generate embeddings for large datasets without managing model infrastructure","Switch between local inference (for privacy/latency) and cloud inference (for scalability) without code changes","Batch embed millions of documents efficiently with automatic request queuing and retry logic"],"best_for":["Teams building RAG systems who want flexibility between local and cloud inference","Organizations with privacy requirements that need local embedding generation","Data engineering teams processing large datasets with variable compute availability"],"limitations":["Cloud API requires authentication and may have rate limits (typically 100-1000 requests/sec depending on tier)","Local inference requires downloading model weights (768-dim model ~500MB, full model ~2GB)","Batching adds latency for small batch sizes; optimal batch size is 32-256 depending on hardware","No built-in caching across sessions; embeddings must be stored externally for reuse"],"requires":["Python 3.8+","nomic Python package (pip install nomic)","API key for cloud inference (free tier available)","4GB+ RAM for local inference, GPU optional but recommended"],"input_types":["text strings or lists of strings","batched text arrays (numpy, pandas, lists)"],"output_types":["dense float32 vectors","embedding metadata (model version, dimensionality, inference mode)"],"categories":["tool-use-integration","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_4","uri":"capability://memory.knowledge.atlas.interactive.2d.projection.and.visualization.of.embeddings","name":"atlas interactive 2d projection and visualization of embeddings","description":"Transforms high-dimensional embeddings into interactive 2D maps that preserve semantic relationships using dimensionality reduction algorithms (UMAP, t-SNE variants). The implementation creates an AtlasProjection object that maintains the mapping between original embeddings and 2D coordinates, enabling interactive exploration through a web-based UI. Supports dynamic filtering, zooming, and semantic search directly on the visualization.","intents":["Visualize and explore patterns in large embedding datasets to identify clusters and outliers","Create shareable interactive maps of datasets for stakeholder exploration and discovery","Debug embedding quality by visually inspecting semantic relationships and identifying misaligned data"],"best_for":["Data scientists and researchers exploring embedding quality and dataset structure","Product teams creating interactive data exploration interfaces","Teams debugging semantic search or RAG system performance"],"limitations":["2D projection inherently loses information from high-dimensional space; visual clusters may not reflect true semantic similarity","Dimensionality reduction is computationally expensive for datasets >1M points (can take hours)","Interactive UI performance degrades with >100k points on typical hardware","Projection is static after generation; adding new embeddings requires full recomputation"],"requires":["Python 3.8+","nomic package with atlas-client","Web browser for interactive UI (Chrome, Firefox, Safari)","Internet connection for cloud-hosted visualizations"],"input_types":["embedding vectors (numpy arrays, lists)","metadata for each embedding (text, labels, tags)"],"output_types":["interactive 2D map (web UI)","projection coordinates (x, y)","shareable map URLs"],"categories":["memory-knowledge","search-retrieval"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_5","uri":"capability://planning.reasoning.automatic.topic.modeling.and.cluster.discovery.from.embeddings","name":"automatic topic modeling and cluster discovery from embeddings","description":"Analyzes embedding distributions to automatically identify semantic topics and clusters without requiring labeled data. The implementation uses clustering algorithms (HDBSCAN, k-means variants) applied to the embedding space, followed by topic extraction that generates human-readable labels for each cluster. Results are integrated into the Atlas visualization, allowing users to explore topics interactively.","intents":["Automatically discover semantic topics in large document collections without manual labeling","Identify and label clusters of similar content for content organization and discovery","Understand the semantic structure of datasets to inform data curation and quality improvements"],"best_for":["Content teams organizing large document repositories","Researchers analyzing text corpora for thematic patterns","Teams building content recommendation systems based on semantic topics"],"limitations":["Topic quality depends heavily on embedding quality; poor embeddings produce meaningless topics","Automatic topic labeling may produce generic or misleading labels requiring manual review","Clustering algorithms have hyperparameters (min_cluster_size, eps) that require tuning for different datasets","Topics are computed once at map creation time; incremental topic updates not supported"],"requires":["Python 3.8+","nomic package","Pre-computed embeddings for all data points","Sufficient data for meaningful clustering (typically >1000 documents)"],"input_types":["embedding vectors","text content (for label generation)","optional: existing cluster assignments"],"output_types":["cluster assignments (integer cluster IDs)","topic labels (generated text descriptions)","cluster statistics (size, coherence metrics)"],"categories":["planning-reasoning","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_6","uri":"capability://data.processing.analysis.duplicate.detection.and.deduplication.across.embeddings","name":"duplicate detection and deduplication across embeddings","description":"Identifies semantically similar or duplicate documents by analyzing embedding similarity without requiring exact string matching. The implementation computes pairwise similarity matrices (or approximate nearest neighbors for large datasets) and applies threshold-based clustering to group duplicates. Supports both exact duplicates (identical embeddings) and near-duplicates (high cosine similarity).","intents":["Identify and remove duplicate documents from datasets before embedding or indexing","Find near-duplicate content that may represent the same information with minor variations","Merge or consolidate duplicate records in data pipelines"],"best_for":["Data engineering teams cleaning datasets before RAG indexing","Content platforms managing user-generated content with duplicates","Research teams deduplicating training datasets"],"limitations":["Similarity threshold is a hyperparameter requiring manual tuning; no universal threshold works across all domains","Pairwise similarity computation is O(n²) for exact methods; requires approximate methods (LSH, FAISS) for >100k documents","Semantic duplicates may not be detected if embeddings don't capture the relevant similarity dimension","No built-in handling of partial duplicates (e.g., one document is a subset of another)"],"requires":["Python 3.8+","nomic package","Pre-computed embeddings for all documents","Sufficient memory for similarity matrix (8 bytes × n² for n documents)"],"input_types":["embedding vectors","document IDs or metadata","optional: similarity threshold (default 0.95)"],"output_types":["duplicate groups (lists of document IDs)","similarity scores between duplicates","deduplication recommendations"],"categories":["data-processing-analysis","search-retrieval"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_7","uri":"capability://search.retrieval.semantic.vector.search.and.retrieval.from.indexed.datasets","name":"semantic vector search and retrieval from indexed datasets","description":"Enables fast semantic search over indexed embeddings by computing similarity between a query embedding and stored document embeddings. The implementation uses approximate nearest neighbor (ANN) algorithms (FAISS, HNSW) for sub-linear search time on large datasets. Supports filtering by metadata tags and returning top-k results with similarity scores.","intents":["Retrieve semantically similar documents for a given query without keyword matching","Build semantic search interfaces for RAG systems and knowledge bases","Find similar examples or references in large document collections"],"best_for":["RAG system builders implementing semantic retrieval components","Search teams building semantic search features","Teams building recommendation systems based on semantic similarity"],"limitations":["ANN algorithms trade recall for speed; approximate search may miss some relevant results (typically 95-99% recall)","Search quality depends entirely on embedding quality; poor embeddings produce poor results","Metadata filtering reduces search speed; complex filters may require full-dataset scans","Index must be rebuilt when adding new documents; no incremental indexing"],"requires":["Python 3.8+","nomic package with atlas-client","Pre-computed embeddings for all documents","Indexed dataset (created via atlas.map_data())"],"input_types":["query text or embedding vector","optional: metadata filter criteria","optional: top-k parameter (default 10)"],"output_types":["ranked list of document IDs","similarity scores (cosine similarity, 0-1)","document metadata and content"],"categories":["search-retrieval","memory-knowledge"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_8","uri":"capability://automation.workflow.progressive.dataset.building.with.incremental.data.addition","name":"progressive dataset building with incremental data addition","description":"Supports adding data to existing Atlas datasets incrementally without full recomputation. The implementation maintains an AtlasDataset object that can accept new documents, embeddings, and metadata through append operations. New data is indexed and integrated into existing visualizations and indices without requiring full dataset reprocessing.","intents":["Build datasets incrementally as new data becomes available without downtime","Add new documents to existing RAG indices without full reindexing","Update visualizations and topic models as datasets grow"],"best_for":["Teams managing continuously growing datasets (news feeds, user-generated content)","RAG systems that need to add new documents without full reindexing","Research projects where data collection is ongoing"],"limitations":["Topic models and 2D projections are not automatically updated; require manual recomputation","Incremental indexing may have slightly higher per-document overhead than batch indexing","No built-in versioning or rollback; adding incorrect data requires manual deletion","Concurrent writes to the same dataset may cause conflicts; requires external synchronization"],"requires":["Python 3.8+","nomic package","Existing AtlasDataset object","Pre-computed embeddings for new documents"],"input_types":["new text documents","new embedding vectors","metadata for new documents"],"output_types":["updated dataset with new documents indexed","confirmation of successful addition"],"categories":["automation-workflow","data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__cap_9","uri":"capability://data.processing.analysis.metadata.tagging.and.filtering.for.data.organization","name":"metadata tagging and filtering for data organization","description":"Enables attaching arbitrary metadata tags to documents and filtering search results or visualizations by tags. The implementation stores metadata alongside embeddings and supports both single-value tags (e.g., category) and multi-value tags (e.g., keywords). Filtering is applied at query time or visualization time to subset data.","intents":["Organize documents by category, source, date, or other attributes for structured exploration","Filter search results by metadata criteria (e.g., 'only show documents from 2024')","Create views of datasets that show only relevant subsets for different users or use cases"],"best_for":["Content teams organizing large document repositories with multiple dimensions","RAG systems that need to filter results by source or metadata","Multi-tenant systems where different users see different subsets of data"],"limitations":["Filtering by metadata reduces search speed; complex filters may require full-dataset scans","No built-in support for hierarchical tags or tag relationships","Tag values are stored as strings; no type validation or schema enforcement","No full-text search on tag values; only exact matching supported"],"requires":["Python 3.8+","nomic package","Metadata for documents (as dictionaries or structured data)"],"input_types":["metadata dictionaries (key-value pairs)","tag names and values","filter criteria (e.g., {'category': 'news', 'year': 2024})"],"output_types":["filtered document lists","filtered visualizations","tag statistics (count by tag value)"],"categories":["data-processing-analysis","search-retrieval"],"confidence":0.5,"matches":0,"success_rate":0},{"id":"nomic-embed__headline","uri":"capability://data.processing.analysis.open.source.multimodal.embedding.platform","name":"open-source multimodal embedding platform","description":"Nomic Embed is an open-source platform for generating and visualizing high-quality text and multimodal embeddings, enabling users to explore large unstructured datasets with transparency and advanced topic modeling.","intents":["best open-source embedding platform","embedding models for text and images","how to visualize embeddings","multimodal data analysis tools","semantic search solutions for unstructured data"],"best_for":["data scientists","machine learning engineers"],"limitations":["requires familiarity with Python"],"requires":["Python environment"],"input_types":["text","images"],"output_types":["embeddings","visualizations"],"categories":["data-processing-analysis"],"confidence":0.5,"matches":0,"success_rate":0}],"trust":{"score":58,"verified":false,"data_access_risk":"low","permissions":["Python 3.8+","PyTorch 1.9+ for model inference","GPU recommended for batch embedding generation (CPU inference ~10-50x slower)","PyTorch 1.9+","PIL/Pillow for image preprocessing","GPU strongly recommended (CPU image encoding ~20-50x slower)","Python 3.8+ (for creating maps)","Web browser for viewing (Chrome, Firefox, Safari)","Nomic account for sharing and permissions","PyTorch Lightning 1.5+"],"failure_modes":["Matryoshka training adds complexity to fine-tuning workflows compared to fixed-dimension models","Quality degradation increases at lower dimensionalities; 128-dim embeddings may lose semantic precision for nuanced queries","No built-in adaptive selection mechanism — applications must implement their own logic to choose dimensionality per query","Image encoding adds 50-200ms per image depending on resolution and hardware","Alignment quality depends on training data diversity; performance may degrade on domain-specific images not well-represented in training set","No built-in support for video or 3D data; only static images","Cross-modal retrieval performance typically 5-15% lower than single-modality search due to alignment trade-offs","Web UI performance degrades with >100k data points; may require data sampling for large datasets","Sharing permissions are managed through Nomic platform; no integration with enterprise SSO","Collaborative annotations are not version-controlled; conflicts possible with concurrent edits","builder identity is not verified yet","no observed match outcomes yet"],"rank_breakdown":{"adoption":0.7,"quality":0.9,"ecosystem":0.6,"match_graph":0.25,"freshness":0.52,"weights":{"adoption":0.3,"quality":0.2,"ecosystem":0.15,"match_graph":0.3,"freshness":0.05}},"observed_outcomes":{"matches":0,"success_rate":0,"avg_confidence":0,"top_intents":[],"last_matched_at":null},"maintenance":{"status":"active","updated_at":"2026-06-17T09:51:04.693Z","last_scraped_at":null,"last_commit":null},"community":{"stars":null,"forks":null,"weekly_downloads":null,"model_downloads":null,"model_likes":null}},"distribution":{"claim_url":"https://unfragile.ai/submit?claim=nomic-embed","compare_url":"https://unfragile.ai/compare?artifact=nomic-embed"}},"signature":"grTME/h2mV/wglwzyq2JaS8D4XcgDizf+jBTGnkETuU+QMx4Lz1okaUahgspGSX88oljxTsiHnEDIlL7Khu5BA==","signedAt":"2026-06-21T15:05:34.396Z","signedBy":"unfragile.ai","version":1},"_links":{"self":"https://unfragile.ai/api/v1/passport/nomic-embed","artifact":"https://unfragile.ai/nomic-embed","verify":"https://unfragile.ai/api/v1/verify?slug=nomic-embed","publicKey":"https://unfragile.ai/api/v1/trust-passport-public-key","spec":"https://unfragile.ai/trust","schema":"https://unfragile.ai/schema.json","docs":"https://unfragile.ai/docs"}}