# Vmax and Modal use self-play to train agents on the Modal SDK

> Original title: Vmax is building Campaign, a platform for RL task generation using self-play.

- Source: Modal
- Published: 2026-11-10T14:36:44.000Z
- Bad Influence: https://badinfluence.news/items/mr8c6kbj6nj147d6bhuw381od
- Original: https://x.com/modal/status/2120163753920504104

## Summary

Modal says it worked with Vmax to teach agents to use the Modal SDK by generating tasks that target their failure modes. The tasks translate Python code into working Modal workloads, and Modal's elastic compute runs and checks the solutions, including GPU calculations and inference endpoints that reuse state across requests.

## Post

Vmax is building Campaign, a platform for RL task generation using self-play.

We worked with them to teach agents to use the Modal SDK, translating Python code into working Modal workloads by generating tasks that target their failure modes. 

Modal's elastic compute lets them run and check solutions, from GPU calculations to inference endpoints that reuse state across requests.

## Quoted post @VmaxAI

> 1/ As a teaser for Campaign, Vmax’s infrastructure for automated task generation, we are sharing work we did earlier this year for @modal using self-play to generate tasks that target weaknesses of models when using the Modal-SDK.

https://x.com/VmaxAI/status/2120154523008368773