NVDA 145.98 ▼3.30%GOOGL 272.66 ▼3.46%MSFT 498.30 ▼1.78%AMD 489.19 ▼7.72%INTC 55.72 ▲1.92%TSMC 352.09 ▼2.90%AMZN 257.37 ▼3.87%META 596.22 ▼3.27%AAPL 349.21 ▼0.92%PLTR 153.68 ▼1.91%
Markets at last close

Apps

Octen expands agent search with fast retrieval and flat-rate pricing

·1 min read

Octen, founded by former Alibaba Cloud AI search lead Kuan Zou, is expanding its search service for AI agents with its Pulse retrieval model, unlimited-search subscriptions and image search. The company aims to make repeated information checks faster and cheaper for developers building assistants that rely on outside sources.

Pulse uses generative retrieval to produce identifiers mapped to existing URLs rather than repeatedly ranking pages in a conventional index. Its lite mode reports a median server-side processing time of about 10 milliseconds, excluding network travel, content fetching and answer generation. Company research suggests strengths in finding familiar destinations, while embedding-based retrieval retains an advantage on infrequently visited pages.

The $20 Standard subscription includes unlimited Web Search and 1,000 image searches monthly. These are personal-account plans whose requests can queue during busy periods, not guarantees of production throughput. Content retrieval and other APIs can incur additional charges, while reserved capacity is sold separately. Independent testing found Octen Search competitive on speed and total task cost, although some rivals scored higher on answer quality. Those results do not validate Pulse’s latency claim.

Self-hosting also carries restrictions: downloadable model weights and the identifier-to-URL table prohibit commercial use without separate licensing. For enterprise buyers, retrieval speed must therefore be weighed against evidence quality, coverage, total spending and deployment rights.

Originally reported by venturebeat.comRead the source →
Related coverage