Open navigation menu

Haystack

deepset

Open Source

LLM pipeline and retrieval orchestration framework rebranded toward the Haystack Enterprise Platform story.

Curated by Tiberiu ArvaVerified

Apache 2.0Version 2.31.0Released 2026-07-0825,865

Strengths

  • Mature pipeline patterns
  • Strong retrieval orientation
  • Production-oriented evaluation tooling

Practitioner note

Strong option when retrieval pipelines and modular backend orchestration matter more than no-code UX or broad business-user automation.

Governance posture

Reviewed 2026-05-26
YesData residency

Python library runs in-process on customer infrastructure; no vendor data handling.

Source
YesDeployment model

OSS library self-hosted (Docker/K8s/VMs); deepset offers a managed Enterprise Platform. (self-hosted, on-prem)

Source
N/AAudit logging

OSS library has no built-in audit log; operators integrate their own.

PartialSOC 2

OSS library N/A; the deepset-managed Haystack Enterprise Platform holds SOC 2.

Source
PartialISO 27001

OSS library N/A; the deepset Enterprise Platform holds ISO 27001.

Source
UnknownISO 42001

Not mentioned in Haystack/deepset public compliance documentation.

N/AEU AI Act

OSS pipeline framework, not an AI model/provider. (role: not-applicable)

YesLicense risk

Apache-2.0 — permissive, OSI-approved. (low)

Source

Change history

Source-backed and auto-detected events for this tool, newest first.

  1. ReleaseHigh impact

    Haystack 2.31.0 began moving components out of core into dedicated packages ahead of 3.0.

    Release v2.31.0 · deepset-ai/haystack
  2. Release

    Haystack 2.30.2 fixed Agent exit behavior so invalid or discarded tool calls no longer cause premature termination under the default text exit condition, allowing the agent loop to recover on the next iteration.

    Release v2.30.2 · deepset-ai/haystack
  3. ReleaseHigh impact

    Haystack 2.28.0 introduced breaking API changes around request exception handling, LLM initialization, and agent invocation signatures, with new runtime State injection for tools.

    Release v2.28.0 · deepset-ai/haystack · GitHub
  4. Release
    Haystack simplifies multi-input pipelines with automatic list joining

    Haystack's April 1 release highlighted automatic list joining inside pipelines, simplifying multi-input workflow construction without extra glue components.

    Releases · deepset-ai/haystack

Explore the category

Compare this tool against the rest of its category and the cloud platform foundation layer.