Skip to content

Grumpy license FAQ

Grumpy is licensed under the Business Source License 1.1 (BSL 1.1) by Imaginary Biolabs GmbH. This FAQ explains common situations in plain language. It is not legal advice; when in doubt, read the LICENSE and contact licensing.

Summary: You may use Grumpy freely for development and for most internal production work. You may not offer Grumpy — or products built primarily around the restricted categories below — to third parties in commercial competition with Imaginary without a separate written license.

When does Grumpy become open source?

Each version converts to Apache License 2.0 on the earlier of:

  1. 2030-12-31 (the Change Date in LICENSE), or
  2. The fourth anniversary of that version’s first public distribution under BSL.

So the first public release (for example on PyPI) may become Apache around four years after launch, even if that is before 2030-12-31.

What production use is allowed without a commercial license?

The Additional Use Grant allows production use except when you offer the Licensed Work (or a substantial portion of it) to third parties as part of a product or service whose primary purpose is any of:

Category Examples (not exhaustive)
(i) Ragged array library A general-purpose numerical or columnar library for ragged/nested scientific data, distributed or hosted for others
(ii) Bio-ML platform / registries Benchmark registry, model marketplace, dataset registry, or similar platform products
(iii) Integrated biology environment An agent or environment that orchestrates biology tools as a product for others
(iv) Fabric-like framework Composable ML pipelines on ragged data, YAML/JSON-driven datasets/benchmarks/metrics/transforms/models/layers, and an integrated biomolecular evaluation engine — including open-source or hosted alternatives to Imaginary’s Fabric product

For commercial entities, those restrictions apply when the offering is in commercial competition with Imaginary or its platform products.

Non-profit academic or research institutions and individuals doing personal non-commercial research may use Grumpy in internal production without the commercial-competition test — but they still may not publish, distribute, or host products falling under (i)–(iv) for third parties without a written license.

Common scenarios

Can I use Grumpy in notebooks, coursework, and papers?

Yes. Non-production and internal production use for research is allowed. Cite Grumpy as you would any dependency.

Can my university lab run HPC training pipelines on Grumpy internally?

Yes, if the lab is a non-profit academic or research institution and use stays internal (cluster jobs, shared lab storage, collaborators inside the project — not a product offered to outside customers).

Can I publish my training code on GitHub?

Usually yes, if the repo is application or experiment code (a model, a benchmark run, analysis scripts) and not primarily a ragged-array library, registry, integrated biology agent, or Fabric-like framework offered to third parties.

If the repo’s main purpose is one of categories (i)–(iv), you need a written license before distributing it — including via public GitHub, PyPI, or a hosted service.

Can I open-source a new ragged-array library built on Grumpy?

Not without a written license if that library is offered to third parties and its primary purpose is category (i) — even if you wrap or extend Grumpy rather than copying it wholesale.

For internal or non-production experimentation, BSL allows modification and redistribution subject to the license terms.

Can my biotech company use Grumpy in drug-discovery pipelines?

Yes for internal production, as long as you are not offering to third parties a product or service whose primary purpose is (i)–(iv) in commercial competition with Imaginary.

Using Grumpy inside proprietary pipelines, models, and analyses is intended to be permitted.

We are building a vertical SaaS (not a general ML platform). Is that OK?

Often yes. Restrictions target products whose primary purpose is a general ragged library, bio-ML platform/registry, integrated biology environment, or Fabric-like framework — not every application that happens to use arrays internally.

If your product’s main value is something else (assay analysis, a specific therapeutic program, a specialized design tool), you are likely outside (i)–(iv). Edge cases should be confirmed with licensing.

What counts as “substantially similar to Fabric”?

Fabric (Imaginary’s ML pipeline layer on Grumpy) is described in category (iv): standardized biomolecular data processing, training, and evaluation on ragged arrays, with composable transform pipelines and config-driven (YAML/JSON) definitions of datasets, benchmarks, metrics, transforms, models, or layers, plus an integrated evaluation engine — whether open source, hosted, or otherwise made available to third parties.

A thin wrapper around PyTorch/JAX for one benchmark is unlikely to qualify. A reusable, config-driven bio-ML stack meant for others to build on likely would.

Can a public consortium (EBI-style) host a benchmark runner on Grumpy?

Depends on primary purpose. A public service whose main role is a benchmark registry or Fabric-like evaluation platform for third parties falls under (ii) or (iv) and requires a written license.

Internal or member-only infrastructure, or publishing results rather than the platform itself, may be fine — confirm for your deployment model.

I’m spinning out a company from academia. What changes?

Internal academic use may have been fine under the carve-out. Once you offer a product or service to customers or the public, normal commercial rules apply — including the commercial-competition test and the (i)–(iv) categories.

Plan for a commercial license if your company’s product is platform-shaped.

Can I fork Grumpy after the Change Date?

Yes. On the Change Date (or four-year anniversary, whichever is first) for a given version, that version is under Apache 2.0, including the right to use it in commercial products without the BSL Additional Use Grant restrictions.

Commercial and partnership licenses

Imaginary offers written licenses for uses outside the Additional Use Grant — for example shipping a Grumpy-based library, a Fabric alternative, or a hosted bio-ML platform.

Contact licensing with a short description of your organization, use case, and whether you plan to distribute or host software for third parties.

Compliance

  • You must retain the LICENSE text with copies and derivatives of Grumpy.
  • Use in violation of the license terminates your rights automatically.

Next: Home