---
title: DeepGym documentation
description: RL environments with verifiable rewards for coding and computer-use agents. Run model output in a sandbox, score it with a verifier, and train on the result.
sidebar:
  label: Overview
---

DeepGym runs model output in a sandbox, checks it with a verifier and returns a score with a per-test breakdown. The score is the reward: it plugs into TRL, verl, OpenRLHF, Axolotl or your own GRPO, DAPO or PPO loop.

For agents, one sandbox stays alive for the whole episode, either a shell or a full desktop. The agent acts step by step, and the verifier scores the state it leaves behind.

```bash
pip install deepgym
```

## How a run works

1. **The model answers**

    A function, a patch or a shell script. An agent instead acts one step at a time in a live sandbox.

2. **DeepGym runs the answer in a sandbox**

    Locally while you develop, or in a Docker, Apple container, Daytona, Modal or E2B sandbox.

3. **The verifier scores it**

    A score from 0 to 1 and a pass flag, plus a result for each test case when the verifier reports them.

4. **Your trainer gets the reward**

    Through the TRL, verl, OpenRLHF or Axolotl adapters, or your own loop. A run that failed for reasons outside the model's control gets no verdict, so it can be kept out of training instead of scored 0. See [the reward contract](/docs/training/reward-contract).

## Start here

**[Quickstart](/docs/quickstart)**

Score a solution, score a batch for GRPO and run the QA gate in a few minutes.

**[Core concepts](/docs/concepts)**

Environments, verifiers, sandboxes, scores and verdicts.

## Explore

**[Environments](/docs/environments/built-ins)**

31 built-ins, benchmark importers, Harbor tasks and computer use.

**[Authoring](/docs/authoring/new-check-qa)**

Scaffold, check, QA-gate and approve your own environments and verifiers.

**[Rewards and training](/docs/training/reward-contract)**

The reward contract and the TRL, verl, OpenRLHF, Axolotl and lm-eval adapters.

**[Traces](/docs/traces)**

Record graded episodes with provenance and export them as training data.

**[Sandboxes](/docs/sandboxes)**

Docker, Apple container, Daytona, Modal and E2B behind one API.

**[CLI reference](/docs/cli)**

Every `deepgym` command and its main flags.

## License

[PolyForm Noncommercial License 1.0.0](https://github.com/DeepGym/deepgym/blob/main/LICENSE). Commercial use needs a license: email [contact@deepgym.io](mailto:contact@deepgym.io).