# postgresml/postgresml

Postgres with GPUs for ML/AI apps.

Repository: https://github.com/postgresml/postgresml
Canonical: https://ross.abutalabs.com/products/postgresml
Homepage: https://postgresml.org
Language: Rust
License: MIT
License Family: permissive
Topics: ml, machine-learning, ai, ann, artificial-intelligence, classification, embeddings, knn, llm, postgres, rag, regression, sql, vector-database, approximate-nearest-neighbor-search, clustering, forecasting
Last push: 2025-07-01T12:26:02+00:00

## Health v2 (maintenance only)
Score: 36/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 29, release rhythm 8, longevity 100
- inputs: {"age_days": 1605, "days_push": 428, "days_rel": 594, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6821, forks 363 (observed 2026-08-28T04:09:48.830068+00:00)

## What it is
PostgresML is a PostgreSQL extension that brings machine learning and AI inference directly into the database, with GPU acceleration and pgvector-based vector search. It lets you run LLMs, embeddings, RAG pipelines, and supervised learning via SQL without moving data to separate ML services.

## Use cases
- build a RAG chatbot without a separate vector database
- generate embeddings for text stored in Postgres using SQL
- run LLM inference and text generation inside the database
- train and deploy classification and regression models on data in Postgres
- improve search results with semantic vector search and ranking
- fine-tune Hugging Face models on your own data
- consolidate vector DB, embedding service, and LLM API into one database

## When to choose
- your data already lives in Postgres and you want ML/AI close to it
- you want to simplify an AI stack by replacing separate vector DB, embedding, and inference services
- you need GPU-accelerated inference with lower latency than HTTP-based model serving
- you prefer SQL as the interface for ML workflows
- data privacy requires keeping models and data together

## When to avoid
- you need a lightweight standalone vector store without a full Postgres deployment
- your ML workloads are better served by specialized frameworks like PyTorch or dedicated serving stacks
- you cannot run GPU-backed Postgres infrastructure and lack cloud budget
- you need non-Postgres databases or multi-database ML orchestration

## Facets
- artifact type: library
- maturity: active
- function: vector-database, machine-learning, llm-inference, rag, search-engine, nlp
- domain: machine-learning, large-language-models, databases, artificial-intelligence
- platform: self-hosted, cloud, python, rust
- tags: postgres-extension, pgvector, huggingface, gpu-acceleration, in-database-ml, sql, embeddings, vector-search, fine-tuning, retrieval-augmented-generation, search, natural-language-processing, linux, docker, gpu

## Member repositories
- postgresml/postgresml (main) score 36

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:48.830068+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:41:55.205166+00:00, confidence not recorded.
  - readme: https://github.com/postgresml/postgresml (fetched 2026-08-28T04:09:48.830068+00:00, sha 27fde752a1d1)
  - homepage: https://postgresml.org (fetched 2026-08-29T08:38:09.971270+00:00, sha 9ee6d24b48e2)
  - site_page: https://postgresml.org/docs/cloud/overview (fetched 2026-08-29T08:38:09.980660+00:00, sha 3967bf4e27ba)
  - site_page: https://postgresml.org/docs/cloud/enterprise/vpc (fetched 2026-08-29T08:38:09.982604+00:00, sha 8ed00db20956)
  - site_page: https://postgresml.org/docs/open-source/pgml/guides/llms (fetched 2026-08-29T08:38:09.984311+00:00, sha 0dd72baeee4b)
  - site_page: https://postgresml.org/docs/open-source/pgml/guides/embeddings (fetched 2026-08-29T08:38:09.985834+00:00, sha a812cb201bd6)
  - site_page: https://postgresml.org/docs/open-source/pgml/guides/vector-database (fetched 2026-08-29T08:38:09.987479+00:00, sha d13d6fff9649)
  - site_page: https://postgresml.org/docs/open-source/pgml/guides/supervised-learning (fetched 2026-08-29T08:38:09.989151+00:00, sha fe335e49b8e8)
  - site_page: https://postgresml.org/docs/open-source/pgml/guides/improve-search-results-with-machine-learning (fetched 2026-08-29T08:38:09.990916+00:00, sha d5bb22bbea47)
  - site_page: https://postgresml.org/pricing (fetched 2026-08-29T08:38:09.993181+00:00, sha f50ecf3015e3)
- Data as of 2026-08-30T08:39:29.467469+00:00.