Multilingual legal files
AI Models & Assistants / Knowledge library
text-embedding-3-large
Most capable embedding model
Why learn text-embedding-3-large?
Offers the documented OpenAI model capabilities through hosted.
The problem it solves
Provides model inference for supported text inputs.
Where it came from
text-embedding-3-large: Most capable embedding model Published aliases: text-embedding-3-large. Input: text. Output: text. Checked 2026-10-13. Model capabilities are distinct from tools available in a particular agent.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Explain what text-embedding-3-large does.
- Identify a use case for text-embedding-3-large.
- Describe a limitation of text-embedding-3-large.
- Name a compatible tool for text-embedding-3-large.
- Compare text-embedding-3-large against an alternative.
Documentation and sources
- Primary source / model card (developers.openai.com)
Catalog research notes
- See the AI Workbench for source dates, pricing evidence and configuration details.
Connect the ideas
Built on
Real World Example
Global Legal Discovery Optimization
International law firms managing massive multilingual document repositories for cross-border litigation.
The business problem
Traditional keyword-based search fails to identify semantically related evidence across different languages, leading to missed critical documents and increased manual review costs.
How this topic solves it
By utilizing text-embedding-3-large, firms can convert multilingual legal documents into high-dimensional numerical vectors. This allows the system to measure the relatedness between documents regardless of the source language, enabling semantic retrieval that surfaces relevant evidence based on conceptual meaning rather than exact keyword matching.
- Pre-process raw legal documents by cleaning and normalizing text across various languages.
- Send the processed text to the v1/embeddings endpoint specifying text-embedding-3-large to generate vector representations.
- Store the resulting high-dimensional vectors in a specialized vector database for efficient similarity search.
- Execute semantic queries by embedding search terms and performing nearest-neighbor lookups against the stored document vectors.
The flow shows how raw text is transformed into searchable vectors.
text-embedding-3-large
Storage for embeddings
Semantic search interface
- Raw Docs → sends text → Embedding API
- Embedding API → stores vectors → Vector DB
- Query Engine → performs similarity search → Vector DB
Implementation example
This code snippet demonstrates calling the v1/embeddings endpoint to transform text into a numerical vector using the specified model.
import openai
# Generate embedding for a legal document
response = openai.Embedding.create(
input="Evidence text here",
model="text-embedding-3-large" # Specifies the most capable model
)
vector = response['data'][0]['embedding']Expected business outcome
Legal teams experience a significant improvement in discovery precision, as the system surfaces conceptually relevant documents that would have been ignored by standard keyword searches.
Limitations and checks
The model produces high-dimensional vectors that require significant storage and computational resources for similarity calculations; developers should monitor latency during large-scale retrieval operations to ensure performance remains within acceptable bounds.