series is the latest generation of state-of-the-art text embedding models from , released in early 2026 . This family features an industry-first shared embedding space
: Optimized for high throughput and low cost, making it ideal for large-scale production queries. voyage-4-nano open-weight model available on Hugging Face for local development and prototyping. Key Technical Features Shared Embedding Space
Use Voyage 4 where length and fidelity matter. For trivial retrieval, reach for a smaller tool. voyage 4
Voyage 4 trails slightly behind Google’s proprietary Gecko on general MTEB but crushes it on long-context retrieval, where Gecko is not competitive.
The standout feature of is the introduction of shared embedding spaces . Historically, switching between a "large" model for accuracy and a "lite" model for speed required re-embedding your entire dataset—a process that is both time-consuming and expensive. series is the latest generation of state-of-the-art text
is a series of state-of-the-art text embedding models released by Voyage AI (now part of MongoDB ) designed for high-accuracy AI retrieval and search. Key Features
Imagine an AI coding assistant that remembers your architectural decisions from last week. Or a travel agent bot that recalls your dietary restrictions from a conversation three months ago. Voyage 4 enables by efficiently embedding, storing, and retrieving entire session histories. Key Technical Features Shared Embedding Space Use Voyage
Whether you are a solo developer hacking on a weekend project or a CTO overseeing a Fortune 500 knowledge base, implementing Voyage 4 will be the single highest-leverage change to your retrieval pipeline in 2025.
In the vast landscape of mobile gaming, where high-octane arcade racers and simplistic drag strips dominate the app stores, a unique gem emerged from the Russian development community. , developed by existage, is not just a game; it is a digital love letter to the Russian countryside, a simulator that prioritizes the journey over the destination, and a test of patience that has garnered a cult following worldwide.
: Voyage 4 is priced at $0.06 per million tokens (input) and $0.12 per million tokens (output for reranking). This is competitive with OpenAI’s text-embedding-3-large ($0.13/1M tokens), but superior for long-context tasks.
: A balanced model that provides high-fidelity retrieval quality with the efficiency of a mid-sized model. voyage-4-lite