Milvus is an open-source vector database for GenAI applications that runs similarity search over billions of vectors · milvus.io ↗
The standing depends on who is buying. Every model is asked the same questions for each company size.
pgvector is first at 35%.
Qdrant is first at 34%. 0 of 14 models made it the first choice when asked directly.
Pinecone is first at 45%.
“If you have a platform/DevOps team: Evaluate Milvus or Qdrant self-hosted for cost control at scale”Kimi K2 · Scale-constrained question · first choice
“Open-source, battle-tested at scale (used by big tech), and offers managed cloud (Zilliz Cloud).”MiniMax M2.5 · Direct question · alternative
“Expect a large distributed deployment or need extensive index choices? Evaluate Milvus.”GPT-6 Luna · Comparative question · alternative
“Milvus: Best for large-scale AI platforms (recommendation systems, image/video search).”MiniMax M2.5 · Comparative question · alternative
“Milvus for small or moderate workloads: ... it can require substantially more memory and operational complexity than lighter alternatives”Perplexity Sonar · Negative question · warning
“If you need a more scalable, distributed solution later, but it's overkill for early stages.”GLM 4.7 FlashX · Budget-constrained question · warning
“Handles massive scale (10B+ vectors) but higher memory use (up to 2.6GB in benchmarks)”Grok 4.20 · Budget-constrained question · warning
“Milvus if you don’t hard-code secure defaults and audit the example Compose stack.”GLM 4.7 FlashX · Negative question · warning
The products named in the same answers as Milvus, over the 114 answers that named it, and how often each was the first choice instead.
| Product | Named together | Picked instead | |
|---|---|---|---|
| 106 | 31 | Compare → | |
| 99 | 25 | Compare → | |
| 92 | 7 | Compare → | |
| 88 | 24 | Compare → | |
| 66 | 2 | View → | |
| 23 | 1 | View → | |
| 22 | 0 | View → | |
| 20 | 0 | View → | |
| 17 | 1 | Compare → | |
| 16 | 0 | View → |
What the models named instead of Milvus →
A side-by-side page exists where both products are among a category's top eight.
Stated on Milvus's own site, not checked by Gane.
Fourteen AI models from 12 labs are asked the same buying questions six ways, for three company sizes. Standings refresh every edition.
Gane sells AI visibility software. No company can pay to change a result.
Read the method →Fourteen of the fourteen models returned a source list in this edition: five return the pages they cited and nine the search results they consulted. These counts come from 96 of the 114 answers that named Milvus and are not a share of its labels.
211 of the 236 domain citations in answers naming Milvus came from a third party's page.
Pages are listed as the models cited them.
| Model | First choice | Alternative | Mention | Warning | Labels |
|---|---|---|---|---|---|
| Claude Haiku 4.5 | 0 | 1 | 2 | 0 | 3 |
| GPT-5.4 mini | 0 | 2 | 0 | 1 | 3 |
| Gemini 3.5 Flash | 0 | 1 | 1 | 1 | 3 |
| Perplexity Sonar | 0 | 1 | 0 | 1 | 2 |
| Grok 4.20 | 0 | 3 | 0 | 1 | 4 |
| Mistral Small | 0 | 0 | 2 | 0 | 2 |
| DeepSeek V4 Flash | 0 | 1 | 1 | 1 | 3 |
| Llama 4 Maverick | 0 | 2 | 1 | 0 | 3 |
| Qwen 3.7 Flash | 0 | 2 | 1 | 0 | 3 |
| Kimi K2 | 1 | 1 | 0 | 1 | 3 |
| GLM 4.7 FlashX | 0 | 2 | 0 | 2 | 4 |
| MiniMax M2.5 | 0 | 3 | 1 | 0 | 4 |
| GPT-6 Luna | 0 | 2 | 0 | 0 | 2 |
| Muse Glimmer 30B | 0 | 4 | 0 | 0 | 4 |
| Kind | Pages | Last 90 days | 2025-10 to 2026-09 | Latest | Categories named |
|---|---|---|---|---|---|
| Glossary or explainer | 1,968 | undated | ETL, Vector databases | ||
| Blog | 571 | undated | Vector databases, Object storage | ||
| Comparison | 95 | undated | Vector databases | ||
| News or press | 12 | undated | |||
| Template or tool | 8 | undated | |||
| Conference or event | 7 | undated | |||
| Webinar or virtual event | 4 | undated | |||
| Report or ebook | 3 | undated | |||
| Case study | 1 | undated |
Every page milvus.io exposes, subdomains included. Kind is read from the address and title. The last 90 days, the latest date and the twelve months count pages by when they were published, from the site's feeds, a date in the address, or the page's own publication date, read from up to a hundred of its most recently changed pages; a page that says only when it last changed is counted in its kind but not in when, so the recent counts are a floor, and a kind none of whose pages gives a publication date reads undated. An event counts as online when its address or title says so (webinar, on demand, virtual or online summit); a conference, summit, trade show, expo or roadshow that does not say so is counted as a conference or event, which on a vendor's site is mostly in person. Read September 29, 2026.