# How OpenEvidence uses Zilliz

**\< 10ms** Search latency with high recall

As published on [zilliz.com](https://zilliz.com/customers/openevidence). Captured by usedby on 2026-10-06.

- Company: [OpenEvidence](https://www.usedby.ai/companies/openevidence.md)
- Tool: [Zilliz](https://www.usedby.ai/tools/zilliz.md)
- Teams: Engineering

## What the story says

OpenEvidence runs an AI clinical decision tool that gives clinicians evidence-based answers from medical sources. It uses Zilliz Cloud as the vector search layer for retrieval, with a private, HIPAA-compliant data path.

Summary written by usedby from the source page, in English. The figures are those of Zilliz and OpenEvidence, not ours.

> Zilliz Cloud consistently delivers sub-10ms latency with 99%+ recall. Each user query triggers multiple vector searches, yet responses return fast enough that the latency is not visible to clinicians.

- **4,000 QPS**: “The system supports roughly 4,000 QPS and serves more than 800,000 clinicians.”
- **25 million**: “In March 2026 alone, OpenEvidence supported 25 million clinical consultations, and the number is still growing rapidly.”
- **99%**: “With sophisticated quantization and refined strategy, Zilliz Cloud achieves 99% recall with an aggressively compressed index to serve high-QPS low-latency workload, unlike competitors that trade accuracy for speed.”

> Zilliz Cloud has helped us create a strong foundation behind the scenes as we continue to grow and serve hundreds of thousands of clinicians.
>
> Jagath Kumar, Head of Performance Engineering

> We believe AI is becoming a meaningful support layer for physicians, but the experience has to feel trustworthy, reliable, and seamless. Building that kind of product requires a strong foundation behind the scenes. Zilliz Cloud has helped us create that foundation as we continue to grow and serve hundreds of thousands of clinicians.
>
> Jagath Kumar, Head of Performance Engineering at OpenEvidence

## What usedby checked

We compared the story with its live page on 2026-10-06.

- Checked: the figure \< 10ms is on the page; its label is our wording.
- Checked: the passage quoted above is copied word for word from the page, near the name of OpenEvidence.
- Checked: OpenEvidence uses Zilliz. Confirmed line. Latest check across sources: 2026-10-06.
- Not checked: the result itself. We quote it; we did not measure it.

---
Source: https://www.usedby.ai/case-studies/openevidence-zilliz · How we check: https://www.usedby.ai/methodology