DeerFlow

What DeerFlow is

DeerFlow combines a workspace interface with a lead agent, delegated agents, skills, memory and execution tools. The current 2.x project aims at tasks that take several steps and produce files or other artifacts, rather than only returning a short conversational answer.

The preserved 1.x branch shows its earlier research-oriented design and includes actual report examples. Those examples remain useful for understanding the intended output, but they are historical evidence. The 2.x rewrite should have its own evaluation before someone relies on it for an important workflow.

How it fits into a workflow

Begin with one research task and a clearly specified output, then inspect both the sources and the produced file. Review the selected sandbox and provider configuration before adding external integrations. An artifact should be assessed for factual support and usability rather than accepted because a run finished.

Examples described as illustrative or proposed trials are editorial suggestions. Documented real builds, when available, are linked separately below.

Requirements, cost and limitations

Current local-development documentation specifies Python 3.12 or later and Node.js 22 or later; the Docker route requires Compose 2.24 or later. Its evaluation starting point is 4 virtual CPUs, 8 GB RAM and 20 GB SSD space, with 25 GB disk suggested for Docker development. Those numbers exclude a locally hosted model and are starting points rather than a guarantee for every task.

As checked on 30 October 2026, DeerFlow is MIT-licensed; model/search APIs, optional sandbox services and hosting remain separate costs. Configure the provider and execution mode before loading important files. Local execution is not host isolation, and exposing a Docker socket can give the gateway powerful host-level access. The exact cost of the historical report is not disclosed.

Maintenance and version context

2026-09-24: v2.1.0 release documents sandbox, memory, authorization and workspace changes.

When to choose it and what to compare

DeerFlow fits someone willing to operate a multi-component workspace and review the resulting artifacts. A simpler research package may be easier for a single report. The deciding question is whether the workspace, skills and execution tools justify the additional configuration and maintenance.

Focused source review 2026-10-30: Current repository, license, setup, permissions and published implementation evidence; not independently executed. Documentation review, not hands-on product testing.

Sources and evidence

Product facts are based on the following published sources, checked 2026-10-30. Workflow suggestions and comparisons are editorial interpretation; PrisAIm has not independently benchmarked this product.