TAAFT
Free mode
100% free
Freemium
Free Trial
Deals
Companies
Voyage AI
Models
Voyage context 3

Voyage context 3

By Voyage AI
voyage-context-3 is a contextualized chunk embedding model for RAG and retrieval systems. Unlike standard embedding models that treat chunks in isolation, it processes the full document in a single forward pass and generates a distinct embedding per chunk encoding both local chunk content and coarse-grained document-level context. This eliminates the need for manual context augmentation such as overlapping chunks, metadata prepending, or LLM-based contextual retrieval. It accepts the same chunk inputs as standard embedding models and produces vectors with identical output dimensions, making it a drop-in replacement in existing pipelines without downstream workflow changes. The model supports embedding dimensions of 256, 512, 1024, and 2048 via Matryoshka learning, and quantization levels of float32, int8, uint8, and binary via quantization-aware training. Per-chunk context window is 32,000 tokens; total document context length is 120,000 tokens. Designed for long unstructured documents, cross-chunk reasoning tasks, and high-sensitivity retrieval in domains such as finance, legal, medical, and technical documentation. Accessible via the Voyage AI API.
Text Gen 7
Released: July 23, 2025

Overview

voyage-context-3 is a contextualized chunk embedding model that produces vectors capturing both chunk-level detail and full document context. It processes entire documents in a single pass and generates a distinct embedding per chunk. Supports flexible output dimensions (256, 512, 1024, 2048) via Matryoshka learning and multiple quantization options (float32, int8, uint8, binary). Per-chunk context window is 32K tokens; total document context is 120K tokens.

🔗Embeds🔍Information retrieval🔎Semantic search

About Voyage AI

Voyage AI provides best-in-class embedding models and rerankers for search and retrieval over unstructured data, used to power retrieval-augmented generation (RAG) and AI applications. It offers general-purpose, domain-specific (finance, legal, code) and company-specific fine-tuned models. Founded in 2023 and based in Palo Alto, the company was acquired by MongoDB, Inc. in February 2025 and now operates as a MongoDB subsidiary.

Industry: Artificial Intelligence
Location: Palo Alto, California, US
Website: voyageai.com
View Company Profile

Related Models

  • Gen 7
    Hunyuan-Large
    Hunyuan-Large is Tencent’s flagship large language model in the Hunyuan family. It’s built for advanced reasoning, coding, and multilingual tasks, with long-context support, tool/function calling, and structured (JSON) outputs for enterprise copilots and RAG applications.
    💻Coding🤖ChatGPT🤖Agents📝Content
    Text
    Released 1y ago
  • Gen 7
    Qwen3 Max Instruct
    Qwen3 Max Instruct is the trillion-parameter chat and tool-use version of Qwen3-Max, ranking near the top of LMArena and scoring about 69.6 on SWE-Bench Verified and 74.8 on Tau2 Bench for real-world coding and agent tasks.
    💻Coding💬Chatting🤖Agents🚀Productivity
    Text
    Released 4mo ago
  • Gen 1
    GPT-J
    6B parameter open-source model; improved performance over GPT-Neo.
    Text
    Released 5y ago
Last updated: June 24, 2026
0 AIs selected
Clear selection
#
Name
Task