Vector databases · Rankedpgvector
Strongest at small business, contested with Qdrant. Ranked 4 of 68 for enterprise buyers.
The standing depends on who is buying. Every model is asked the same questions for each company size.
Ranked first, contested: Qdrant has 29%.
Qdrant is first at 34%. 5 of 14 models made it the first choice when asked directly.
Pinecone is first at 45%.
Why models recommend it
“the recommendation is to start with pgvector unless your scale, query patterns, or operational constraints provide specific evidence otherwise”Claude Haiku 4.5 · Scale-constrained question · first choice
“Start simple: Use pgvector (in Postgres) or SQLite with vector extensions for most apps—it's battle-tested, avoids new tech debt”Grok 4.20 · Negative question · first choice
“If your company already uses PostgreSQL for standard application data, pgvector is almost always the most cost-effective choice.”Gemini 3.5 Flash · Budget-constrained question · first choice
“The best vector database for a mid-market B2B company is pgvector... It is the strongest default for mid-market enterprises.”Muse Glimmer 30B · Direct question · first choice
What gives them pause
“pgvector does not natively support GPU acceleration or disk-optimized indexes, and at billion-scale workloads typically require more advanced sharding strategies”Claude Haiku 4.5 · Negative question · warning
“Using vectors beside your application data in Postgres via pgvector is often the safest default until measured retrieval requirements force a dedicated engine.”Muse Glimmer 30B · Negative question · warning
“pgvector for large, high-throughput workloads with selective filters or many tenants... approximate-index filtering can return too few results”GPT-6 Luna · Negative question · warning
“Great for <5M-10M vectors. However, building HNSW indexes on tens of millions of vectors requires huge amounts of RAM”Gemini 3.5 Flash · Negative question · warning
Also on the shortlist
The products named in the same answers as pgvector, over the 192 answers that named it, and how often each was the first choice instead.
| Product | Named together | Picked instead | |
|---|---|---|---|
| 172 | 49 | Compare → | |
| 171 | 44 | Compare → | |
| 139 | 7 | Compare → | |
| 95 | 2 | View → | |
| 88 | 4 | Compare → | |
| 38 | 2 | View → | |
| 32 | 0 | View → | |
| 26 | 0 | View → | |
| 25 | 1 | Compare → | |
| 24 | 2 | View → |
What the models named instead of pgvector →
A side-by-side page exists where both products are among a category's top eight.
How this is measured
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 →Sources returned with these answers
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 172 of the 192 answers that named pgvector and are not a share of its labels.
Domains cited
416 of the 444 domain citations in answers naming pgvector came from a third party's page. Twenty-eight were to the publisher's own pages (the Gane Index), which the models cite like any other source; they are not counted as a third party.
Pages cited
Pages are listed as the models cited them.
By model
| Model | First choice | Alternative | Mention | Warning | Labels |
|---|---|---|---|---|---|
| Claude Haiku 4.5 | 1 | 2 | 2 | 1 | 6 |
| GPT-5.4 mini | 1 | 3 | 0 | 0 | 4 |
| Gemini 3.5 Flash | 2 | 3 | 0 | 1 | 6 |
| Perplexity Sonar | 2 | 3 | 0 | 0 | 5 |
| Grok 4.20 | 2 | 4 | 0 | 0 | 6 |
| Mistral Small | 1 | 1 | 3 | 0 | 5 |
| DeepSeek V4 Flash | 2 | 3 | 0 | 1 | 6 |
| Llama 4 Maverick | 1 | 3 | 1 | 0 | 5 |
| Qwen 3.7 Flash | 2 | 3 | 1 | 0 | 6 |
| Kimi K2 | 2 | 4 | 0 | 0 | 6 |
| GLM 4.7 FlashX | 1 | 1 | 0 | 1 | 3 |
| MiniMax M2.5 | 1 | 2 | 0 | 0 | 3 |
| GPT-6 Luna | 1 | 4 | 0 | 1 | 6 |
| Muse Glimmer 30B | 2 | 3 | 0 | 1 | 6 |