AI ranking
Local / open-source agentic systems
Local and open-source agentic systems from Le Recul's editorial catalogue (OpenClaw, OpenHands, OpenClaude, Aider…). Le Recul Agentic score (0–90 method).
Visual top 10 — Le Recul score
- 01 OpenClaw 87.30
- 02 OpenHands 87.30
- 03 OpenClaude 85.50
- 04 SWE-agent 69.30
- 05 Aider Architect 67.50
- 06 OpenCode 67.50
- 07 Hermes Agent 65.70
Compare models. Tick up to 3 models in the table below, then run the comparison: who wins on which use, by how much, and how solid the measurement is.
| # | Model | Vendor | Score | Assessment | |
|---|---|---|---|---|---|
| 01 | Open-source agent by Peter Steinberger for browser automation with the LLM of your choice. | Peter steinberger | 87.30 / 100 | Le Recul | |
| 02 | Open-source coding agent (formerly OpenDevin), bring your own model. Present in SWE-bench. | All hands ai | 87.30 / 100 | Le Recul | |
| 03 | Open-source agent inspired by Claude Cowork, bring your own model. | Anthropic community | 85.50 / 100 | Le Recul | |
| 04 | Academic agentic scaffold for SWE-bench, bring your own model. | Princeton nlp | 69.30 / 100 | Le Recul | |
| 05 | Aider's architect mode — agentic planning separated from editing. Bring your own model. | Aider | 67.50 / 100 | Le Recul | |
| 06 | Open-source command-line coding agent, an alternative to Claude Code. | Opencode community | 67.50 / 100 | Le Recul | |
| 07 | Nous Research agent built on Hermes 3, open source, bring your own model. | Nous research | 65.70 / 100 | Le Recul |
Indicative score, weighted from public sources. Transparent methodology. See the AI ranking page for the full methodology and the other categories.
Local and open-source agentic systems
This view gathers the open or locally runnable agentic systems from Le Recul's editorial catalogue: the agents whose workings you can inspect, which you can host yourself and whose access you control.
That is a difference of nature, not just of licence. With an open-source agent, you can see what it sends, where, and with which permissions. For a task touching files, credentials or business data, that legibility is often worth more than extra performance.
The trade-off is real: installation, configuration, updates and security are on you. A badly configured open-source agent is riskier than a properly framed hosted one. This view is an editorial selection, with no numerical score, for want of comparable public measurement for these systems.
What this ranking is for
This view helps you spot the open agentic systems that are credible for real use, and understand what they demand in configuration and supervision.
It also helps you ask the right question before installing: what access does the agent get on the machine, and what remains reversible if it gets things wrong.
How to read the ranking
- Real openness: inspectable code, self-hosting possible.
- Control over permissions and access granted on the machine.
- Installation and configuration effort to expect.
- Project activity and seriousness of maintenance.
- Reversibility of the actions the agent takes.
FAQ — Local / open-source agentic systems
Is an open-source agent safer?
It is more legible, which is not the same thing. Safety depends above all on your configuration and the permissions granted, not only on the licence.
Why no numerical score?
Because no reproducible public measurement makes it possible today to compare complete agentic systems seriously. This page owns an editorial selection.
What is the difference with local agentic AI?
This view covers agentic systems — the complete tooling. Local agentic AI refers to the models that can power them.
Do you need technical skills?
Yes, more than with a hosted agent. Installation, configuration and keeping up with updates are part of the real cost.
How often does this page change?
It follows the editorial catalogue, updated as tests and the projects evolve. The date is shown at the top of the page.