AI Business Strategy

Octen brings LLM-style pricing to AI search with a $20 unlimited plan and 10ms retrieval

As AI agents become more autonomous, they face a glaring bottleneck: web search. Every time an agent needs to look up real-time information to ground its answers, it has to hit a search API, wait hundreds of milliseconds, and rack up per-call charges. Over dozens of steps, the latency compounds and the bills climb.

Octen wants to fix both the speed and the economics of agentic search. Today, the company announced the launch of Octen Pulse, a new open-source generative retrieval model that it claims returns text results in just 10 milliseconds. Alongside the new model, Octen is rolling out a $20-per-month subscription offering individual developers unlimited API calls, effectively bringing the predictable ChatGPT-style subscription model to the search API market.

“Every time the internet got faster, a wave of new applications followed, and search for AI works the same way,” said Octen founder and CEO Kuan Zou. “How fast a model can retrieve information decides how fast intelligent products can ship, which is why we’re so aggressive about speed.”

The standard paradigm for AI search is highly sequential: a model finishes a reasoning step, pauses to search the web (which typically takes north of 300 milliseconds via traditional APIs), waits for the results, and then resumes. Octen claims that Pulse is about 30 times faster than most AI search APIs. At 10 milliseconds, a search takes roughly the same amount of time and compute cost as generating a single token. This allows retrieval to run concurrently inside a model’s reasoning process rather than acting as a roadblock.

Because Pulse is open-source and capable of running on-device, it’s also positioned for physical AI applications, like robotics and smart glasses, that require real-time context without the latency of a cloud round-trip.

Speed is only half the equation; cost is the other. In the current per-call pricing landscape, developers often artificially limit how often their agents are allowed to search to keep bills down, which inevitably degrades the quality of the AI’s output. By switching to a flat $20 monthly fee for unlimited calls, Octen is betting that developers will let their agents search as often as necessary.

“For agents that rely on the web, search can sometimes cost more than the model itself,” noted Sid Sharma, Head of GTM at Inception. “What we like about Octen is that it makes search spend easier to predict.”

Pulse is the third tier in Octen’s growing search lineup, joining Octen Fast (the company’s core API, which clocks a median of 62 milliseconds on the SealQA Hard benchmark) and Octen Premier (tailored for specialized business and news queries).

With today’s update, the year-old startup is also expanding into multimodal territory. Octen’s subscription now includes image search, allowing models to pull visual references to ground AI image generation, or enabling agents to compare products using both text and photos in a single step.

Octen has some serious pedigree under the hood. Zou previously led enterprise search at Baidu and AI search at Alibaba Cloud, where he helped build infrastructure designed to handle the massive traffic spikes of Singles’ Day. The broader Octen team includes alumni from Meta, Google, TikTok, and DeepSeek.

The company, backed by Square Peg and Argor, says it has grown to over 5,000 active developers on the platform since reaching general availability in July. As the AI industry shifts its focus from static chatbots to active, web-browsing agents, Octen is hoping its utility-like approach to search will make it the default infrastructure for the next wave of autonomous tools.

Related Articles

Back to top button