GPTSwarm
GPTSwarm is a graph-based framework for LLM agents from the paper "Language Agents as Optimizable Graphs", selected as an oral presentation at ICML 2024. It treats an agent as a graph of operations, and a swarm as a composite graph of agents, so both can be built, executed and then optimised.
The library is organised into modules: swarm.environment for domain-specific operations, agents, tools and tasks; swarm.graph for creating and running agent graphs and swarm graphs; swarm.llm for choosing LLM backends and tracking their cost; swarm.memory for index-based memory; and swarm.optimizer for algorithms that improve agent performance and how the swarm is connected, which is the self-improvement capability the project emphasises.
Install with `pip install gptswarm`. It is best suited to researchers and builders who want to experiment with automatically optimising multi-agent structures rather than to teams looking for a production orchestration product. MIT licensed; most of the README's news dates from 2024, and our data shows the last push on 2026-02-05.
Features
Why choose it
Trade-offs
Maintenance check
- Last commit
- 2026-02-05
- Commits, last 90 days
- 0
- Releases, last 90 days
- 0
- Latest release
- —
- Open issues
- 7
- Recent issues with at least one comment
- —
Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.
Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.
Compatibility
Quick start
Use cases
Alternatives
Comments
No comments yet. Be the first!