Octen, an AI-focused search technology company, has launched Octen Pulse, an open-source generative retrieval model designed to deliver text search results in just 10 milliseconds. The company describes the technology as a search engine built specifically for machines, enabling AI models and agents to retrieve information while processing queries and reasoning through tasks.
Alongside the launch, Octen has introduced a $20-per-month subscription offering unlimited API calls for individual developers. The new pricing model aims to eliminate the cost limitations associated with conventional AI search APIs, allowing developers to integrate frequent search operations into their AI applications without worrying about per-query charges.
The launch also expands Octen's search capabilities with image search, supporting the growing demand for multimodal AI applications. Founded by Kuan Zou, who previously led AI search at Alibaba and enterprise search at Baidu, Octen is positioning its technology as infrastructure for the next generation of intelligent applications.
Octen has launched Pulse, an open-source generative retrieval model that delivers text search results in approximately 10 milliseconds.
Octen Pulse is designed to enable AI agents to search while reasoning, reducing delays associated with conventional search APIs.
The company has introduced a $20 monthly subscription providing individual developers with unlimited API calls.
Octen offers three search methods: Pulse for ultra-fast generative retrieval, Fast for structured search results, and Premier for specialized searches.
The subscription now includes image search, supporting multimodal AI applications that combine text and visual information.
Since announcing general availability in July, Octen has grown to more than 5,000 active developers.
Octen Pulse is a generative retrieval model that returns search results in approximately 10 milliseconds. Unlike conventional search systems that retrieve information through multiple processing stages, Pulse has learned millions of URLs across major websites, allowing a single inference to transform a question directly into a matching result.
According to Octen, each search takes approximately the same time and computational cost as generating a single token. This enables AI models to incorporate information retrieval directly into their reasoning processes rather than treating search as a separate operation that interrupts execution.
The model is open source and can run on-device, extending its potential applications beyond cloud-based AI systems. This capability is particularly relevant to physical AI applications, including robots, smart glasses, and smartphones, where low-latency information retrieval is essential.
By reducing search latency, Octen aims to make information retrieval a continuous part of AI reasoning, helping developers build more responsive intelligent applications.
Octen offers three search methods designed to address different information retrieval requirements across AI applications and developer workflows.
Octen Pulse is designed for applications that require near-instantaneous information retrieval. Its generative retrieval architecture produces matching search results directly from a query, reducing the time required for conventional search operations.
With its 10-millisecond response time and open-source availability, Pulse is intended to support AI agents, on-device applications, and physical AI systems that require fast information access.
Octen Fast serves as the company's core search API, delivering structured search results with low latency.
On the SealQA Hard benchmark, Octen Fast achieves a median response time of 62 milliseconds, which the company reports is more than four times faster than its nearest competitor.
In the Artificial Analysis Search API Benchmark, Octen recorded 0.21 seconds per search query and 15.1 seconds per task, making it the fastest provider in the benchmark cited by the company.
Octen Fast also supports more than one million queries per second, making it suitable for large-scale AI applications that require frequent information retrieval.
Octen Premier is designed for specialized information retrieval requirements, including business search, news search, and place search.
These capabilities are available through the Octen API and its console, allowing developers and AI teams to integrate targeted search functionality into their applications.
One of the key developments accompanying the Octen Pulse launch is a subscription model that provides individual developers with unlimited API calls for $20 per month.
Traditional AI search APIs generally charge developers for every request. Since AI agents may perform dozens of searches while completing a single task, these costs can increase rapidly as application usage grows.
This pricing structure can lead developers to restrict how frequently their agents access external information, potentially affecting the accuracy and quality of generated responses.
Octen's subscription model aims to address this challenge by introducing an approach similar to the subscription pricing commonly used by large language model (LLM) providers. By offering unlimited API calls, the company intends to make search usage more predictable and reduce the need for developers to limit retrieval operations.
Octen also highlighted its cost performance in independent testing conducted by Artificial Analysis. The benchmark evaluates the same model and testing framework across 26 search API products from 13 providers.
According to the September results cited by Octen, its search cost was $9.07 per 1,000 tasks, compared with $30 to $66 for other providers.
Sid Sharma, Head of GTM at Inception, commented on the importance of predictable search costs for AI applications:
“For agents that rely on the web, search can sometimes cost more than the model itself. What we like about Octen is that it makes search spend easier to predict, so developers can focus on when their agents need to look something up and spend less time worrying about the bill,”
Octen's subscription now includes image search, expanding its platform beyond text-based information retrieval.
Image search enables AI applications to retrieve relevant visual references that can support image generation models and improve the accuracy of generated content. It also allows AI agents to combine textual and visual information when responding to user requests.
For example, an AI agent comparing products can retrieve product images through the same service it uses for text search. This allows the application to gather relevant information and visual references within a unified workflow.
The addition of image search reflects the growing importance of multimodal AI systems that process and generate different types of information, including text and images.
Professor Aixin Sun, Associate Dean of the College of Computing and Data Science at Nanyang Technological University, highlighted the importance of information accessibility for AI systems:
“An LLM is like a powerful brain. Its true potential is unlocked when it can access the world's latest information instantly and explore knowledge at scale,” said Professor Aixin Sun, Associate Dean of the College of Computing and Data Science at Nanyang Technological University. “Intelligence becomes far more powerful when information is abundant, current, and immediately available.”
Octen's search infrastructure is being developed by a team with experience building large-scale search and recommendation systems at some of the world's leading technology companies.
Founder and CEO Kuan Zou previously led enterprise search at Baidu and AI search at Alibaba Cloud. During his time at Alibaba, he helped develop search infrastructure capable of handling significant traffic spikes, including those associated with Singles' Day.
At Octen, Zou has focused on developing a search engine specifically designed for machine-driven information retrieval.
The company's team includes professionals with backgrounds at Meta, Google, TikTok, Alibaba, Baidu, DeepSeek, and Xiaohongshu.
Octen aims to prepare search infrastructure for the growing volume of machine-generated queries while making information retrieval more reliable, affordable, and accessible for developers.
The company's approach reflects the increasing demand for search systems that can support AI agents operating continuously and retrieving information as part of complex tasks.
Octen's launch of Pulse and its unlimited API subscription highlight the growing importance of low-latency, cost-efficient search infrastructure in the development of AI applications. By combining generative retrieval, structured search, specialized search capabilities, and image search, the company is working toward making information retrieval an integrated part of AI reasoning and execution.
With more than 5,000 active developers since announcing general availability in July, Octen is expanding its platform to support the evolving requirements of agentic and multimodal AI systems. Developers and AI teams can obtain an API key and run their first queries through the company's website at octen.ai.
About Octen
Octen is the search engine built for machines. Its API gives AI models and agents search across text and images, with text results in 10 milliseconds, and ranks results by relevance to the ground truth rather than predicted clicks. Founder and CEO Kuan Zou previously led AI search at Alibaba Cloud and enterprise search at Baidu. Octen is backed by Square Peg and Argor. The company is headquartered in San Francisco.