Cloud Parts

Airbyte

An ELT platform you can host yourself: hundreds of connectors, each a Docker image speaking one JSON protocol, scheduled by services on Kubernetes.

What it is
ETL & ELT pipelines
Type
Commercial product
Made by
Airbyte, Inc.
License
Elastic License 2.0 (source-available, not OSI open source): the platform since 0.30.0 (September 2021), every connector since July 2025; the Airbyte Protocol, Python CDK and abctl are MIT; Airbyte Cloud is proprietary
Facts checked
6 October 2026
Website Source code

What it is

Airbyte has two halves. The platform is a set of services on Kubernetes: an API server and web UI, a Postgres database that keeps connection settings and job history, Temporal for scheduling, and workers that start each sync as its own pods. Connectors are Docker images that follow the Airbyte Protocol. A source answers spec, check, discover (list the tables or API resources it can read) and read, which streams records plus STATE messages; a destination’s write loads them. The platform saves the last state, so the next sync only reads what changed, and Postgres, MySQL, SQL Server and MongoDB sources can follow the database log (CDC). Most connectors no longer need code: 514 of the 690 in the main repo are “manifest-only”, a YAML file run by a shared image, which is what the no-code Connector Builder writes.

You can run it yourself with the Helm chart on an existing cluster, or with abctl, which creates a one-node kind cluster inside Docker. Airbyte Cloud bills in credits: on Standard ($20 a month with 5 credits, then $5 per credit) a million API rows costs 6 credits and a GB from a database or file costs 4, so roughly $30 and $20. Plus sells prepaid packages from $189 a month for 40 credits. Pro and Enterprise Flex are priced by capacity, in “Data Workers” that each run about three syncs at once, through sales. PyAirbyte runs the same connectors inside a Python process and caches the rows in a local DuckDB file by default.

Why teams pick it

  • A large catalogue: about 600 sources and 80 destinations in the main repo, with CDC for the main databases
  • Connector Builder: point-and-click REST API sources whose YAML manifest you can export and edit
  • The same connectors run self-hosted, in Airbyte Cloud and from Python with PyAirbyte
  • A language-neutral protocol: any program in a Docker image that reads config and prints JSON can be a connector
  • Self-hosted use is free, with no limits on volume, users or connections

Watch out for

  • Not OSI open source: ELv2 forbids selling Airbyte as a hosted service, and since July 2025 that covers every connector too
  • Heavy to self-host: even a laptop install runs Kubernetes (kind), Temporal, Postgres and MinIO; Docker Compose installs were deprecated in 1.0 (2024)
  • Cloud meters database and file sources by the size of the serialized records, which Airbyte says is likely larger than in your database; overage is billed in arrears and syncs keep running
  • Batch only: the fastest Cloud schedule is 1 hour on Standard, 15 minutes on Plus, 5 minutes on Pro
  • Self-hosted releases slowed after 2.0 (October 2025): 2.1 came in April 2026 marked experimental, and Self-Managed Enterprise users are told not to upgrade but to move to Cloud

Alternatives

What all etl & elt pipelines have in common →
Apache NiFiOpen sourceA Java dataflow server you program by drawing a graph of processors in a web UI, with buffered queues, back pressure and a record of where every piece of data went.ArtieCommercialA managed change data capture service that copies every insert, update and delete from production databases into a warehouse, aiming for under a minute of delay.BruinOpen sourceOne Go command-line tool that ingests data, runs SQL and Python transformations and checks the results, with each step a file in your repo.DagsterOpen sourceA Python data orchestrator built around assets: you declare the tables, files and models that should exist, and Dagster works out what to run.dbtOpen sourceSQL `select` statements in git, built into warehouse tables in dependency order, with tests and docs alongside.EstuaryCommercialA managed data-movement platform that captures changes from databases and SaaS sources into collections kept as JSON files in your own bucket, then keeps warehouses and other systems in sync, in real time or on a schedule.FivetranCommercialManaged connectors that copy data from SaaS apps, databases and files into your warehouse or lake, billed by the rows that change each month.MeltanoOpen sourceA command-line ELT tool that runs Singer taps and targets from a `meltano.yml` project kept in git.OLakeOpen sourceAn open-source tool in Go that copies databases, Kafka topics and S3 files into Apache Iceberg tables (or plain Parquet), with full loads, incremental syncs and CDC, and no Kafka, Debezium, Spark or Flink in between.

How it fits with other technologies

Read GitHub issues with PyAirbyte into a local DuckDB cache · python

# pip install airbyte; export GITHUB_TOKEN=...
import airbyte as ab

source = ab.get_source(
    "source-github",
    config={
        "repositories": ["airbytehq/airbyte"],
        "credentials": {"personal_access_token": ab.get_secret("GITHUB_TOKEN")},
    },
    install_if_missing=True,
)
source.check()                                    # the protocol's check command
source.select_streams(["issues", "pull_requests"])
result = source.read()                            # lands in a local DuckDB file

issues = result["issues"].to_pandas()
print(len(issues), "issues")

Self-hosted and Cloud plans#

All plans run the same connectors. List prices checked on 6 October 2026.

PlanWho runs itBillingFastest schedulePrice
Core (self-managed)You, on Kubernetes (Helm or abctl)None; ELv2 licenceSet by youFree
StandardAirbyteCredits by volume1 hour$20/month with 5 credits, then $5 per credit
PlusAirbytePrepaid credit packages15 minutesFrom $189/month for 40 credits; overage $5 per credit
ProAirbyteCapacity (Data Workers)5 minutesQuote
Enterprise FlexAirbyte’s control plane; data plane inside your networkCapacity (Data Workers)5 minutesQuote
  • Credit metering on Standard and Plus: 6 credits per million rows from APIs and custom sources, 4 credits per GB from databases and files.
  • New Cloud accounts get a 14-day trial with 400 credits.

Which part is under which licence#

Platform

ELv2

Moved from MIT to the Elastic License 2.0 with version 0.30.0 on 27 September 2021. Free to run, change and use internally; not to resell as a hosted ELT service.

Connectors

ELv2

About two dozen “strategic” connectors (Postgres, MySQL, Salesforce, Snowflake and others) moved to ELv2 in June 2023; a July 2025 change moved the rest. Earlier contributions keep their original licence.

Protocol, Python CDK and `abctl`

MIT

The message format, the Python connector development kit and the local installer remain MIT, so you can build your own connectors under any licence.

Airbyte Cloud, Enterprise

commercial

Hosted and enterprise editions need a commercial agreement with Airbyte.

See it in a real stack

No architecture uses Airbyte yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2020–2025

Michel Tricot had run integrations at LiveRamp, where his team built thousands of connectors; John Lafleur had founded Anaxi. Interviewing data teams in 2020, they found that every company “had to develop connectors internally, even though they were paying for another product”: hosted tools covered popular sources but not the long tail, and could not be changed. Their answer was an open-source platform where anyone could add a connector.

They started in July 2020 and tagged the first alpha that September, with three sources and two destinations. The community grew it fast: in 2022 about 70% of new connectors came from outside contributors. A $26M Series A in 2021 was followed that December by $150M from Altimeter and Coatue at a $1.5 billion valuation. Airbyte Cloud launched in April 2022, the same month Airbyte bought Grouparoo, an open-source reverse-ETL tool.

To stop larger companies reselling the project as a service, Airbyte moved the platform from MIT to the Elastic License in September 2021, then its key connectors in 2023 and the rest in 2025. Version 1.0 in September 2024 swapped Docker Compose for Kubernetes everywhere, with abctl for single machines.

Why it’s built this way

Connectors as Docker images speaking a JSON protocol
Contributors could write a connector in any language and test it alone, and the platform could run, isolate and upgrade each one the same way.
Declarative YAML connectors and a no-code builder
Most REST APIs differ only in authentication, pagination and record paths. Describing those in a manifest made hundreds of API connectors cheaper to build and maintain than Python code.
The Elastic License instead of MIT
Airbyte wanted to keep self-hosting free while preventing cloud providers and competitors from selling the same software as a managed service.

Timeline

  1. 2020

    Founded in July; first alpha release tagged in September

  2. 2021

    Platform moves to ELv2 at 0.30.0; $150M Series B in December

  3. 2022

    Airbyte Cloud launches and Grouparoo is acquired, both in April

  4. 2023

    No-code Connector Builder; strategic connectors move to ELv2

  5. 2024

    PyAirbyte in February; 1.0 with abctl in September

  6. 2025

    All connectors move to ELv2; Airbyte 2.0 in October

Sources25 · facts checked 6 October 2026

Improve this page on GitHub