StableMRR
P

PostgresML pricing

GPU-backed Postgres database for running embeddings, vector search, and open-source LLMs inside SQL.

Free plan
No
Paid from
$0.60/mo
Plans
3
Free trial
-

What is PostgresML?

PostgresML is a database platform that adds GPU-backed machine learning to Postgres, letting teams run embeddings, vector search, and open-source LLMs directly in SQL or through Python and JavaScript SDKs. It is aimed at developers building RAG, search, and chatbot applications who want vectors, models, and application data in one place instead of separate services. Features include vector indexing with HNSW or IVFFlat, embedding generation, model training and fine-tuning, and inference on open-source models. It is sold as PostgresML Cloud and VPC deployments, with usage-based pricing for models and compute.

Plans and prices

Serverless

-

Usage steps

  • per query hour: $7.50
storage
$0.25/GB per month
queries
$7.50 per hour
  • Pay-per-use
  • Burst GPU capacity
  • Access curated models
  • Support on Discord

Dedicated

-

Usage steps

  • per instance hour: $0.60
  • Committed use discounts
  • Dedicated hardware
  • Use any HuggingFace model
  • Deploy on major cloud providers
  • Dedicated support on Slack or MS Teams

Enterprise

Customcontact sales

  • VPC deployments
  • Multiple GPUs
  • Custom SLAs
  • Premium support and onboarding
  • Priority feature requests

Prices in USD, as shown by postgresml.org. Source: pricing page. Last checked October 30, 2026.

Not sure which PostgresML plan fits? Talk it through with your AI agent

Copy the prompt below and paste it into the agent you already use. It has the plans and prices from this page and asks you the right questions first.

See the prompt
I want to decide whether PostgresML (postgresml.org) is the right tool and which plan to pick.
Below is what the StableMRR pricing directory read from PostgresML's own pricing page (last checked 2026-10-30). It can be out of date, so treat it as a starting point.

Category: Developer tools / Databases and backend
What it does: GPU-backed Postgres database for running embeddings, vector search, and open-source LLMs inside SQL.
Plans:
- Serverless: -. Limits: storage $0.25/GB per month, queries $7.50 per hour. Includes: Pay-per-use, Burst GPU capacity, Access curated models, Support on Discord
- Dedicated: -. Includes: Committed use discounts, Dedicated hardware, Use any HuggingFace model, Deploy on major cloud providers, Dedicated support on Slack or MS Teams
- Enterprise: Contact sales. Includes: VPC deployments, Multiple GPUs, Custom SLAs, Premium support and onboarding, Priority feature requests
Alternatives in the directory: ApertureDB, Avo, Backendless, Caspio
Pricing page: https://postgresml.org/pricing
Directory listing: https://stablemrr.com/p/postgresml-org

First ask me three short questions about my team size, budget and what I need it for. Then recommend one plan, tell me what could make it a bad fit, and compare it with the alternatives. Finish with what I should check on the vendor's site before I pay.

Pricing facts

Pricing model
Usage based
Billing
monthly
Discounts
Committed use discounts

About PostgresML

Founded
2022
Platforms
api
Integrations
PyTorch, TensorFlow, Flax, SciKit-Learn, Hugging Face, Llama, Mistral, XGBoost, LightGBM, CatBoost, Falcon, OpenAI and 12 more

Features

  • Pay-per-use billing
  • Burst GPU capacity
  • Curated models
  • Dedicated hardware
  • VPC deployments
  • Custom SLAs
  • Committed use discounts
  • $100 free usage credits

Price history

No price changes recorded since we started tracking on October 30, 2026.

Alternatives to PostgresML

Newsletter

Get a note when PostgresML or its alternatives change their prices

Price changes and new tools, once or twice a month. No pitches, unsubscribe in one click.

No spam. By subscribing you accept the privacy policy. Unsubscribe any time.

How this page was made. Plans and prices are read automatically from postgresml.org's public pricing page, checked against the page text, and re-checked about every week. Method. Something wrong? Tell us.