#5252nd of 63 in Coding agents and assistants

openevolve

Evolves programs with LLM mutations, scored by your own evaluator

Stars: 7.5k
License: Apache-2.0
Last commit: Oct 2026
Last release: Sep 2026
Language: Python

Missed the latest GitHub check; these facts may be older.

Overview

OpenEvolve takes an initial program and an evaluator function, then repeatedly asks LLMs to rewrite the code. It keeps candidates in a MAP-Elites population split across islands. It works with OpenAI-compatible APIs, local servers such as Ollama, or the Claude Code, Copilot and Codex CLIs as backends. It runs from the command line or as a Python library.

Who it is for: Engineers and researchers optimizing algorithms or kernels against a score

Strengths

MAP-Elites with multiple islands and migration, plus island-selector and population hooks
Works with OpenAI-compatible APIs, Ollama/vLLM, and Claude Code, Copilot or Codex CLIs
Usable as a CLI or a Python library with inline code and callable evaluators
Seeded by default (seed 42); cascade evaluation filters weak programs early

Weaknesses

Needs a user-written evaluator; it does not work on an arbitrary repo by itself
LLM cost is real: README estimates $0.01-0.60 per iteration depending on model
Benchmark claims in the README are not backed by numbers beyond the examples
Quick start sets a Gemini key through OPENAI_API_KEY, which is confusing

What it needs

no GPU
Docker
Needs Python 3.10+, LLM provider (OpenAI-compatible API or Claude Code/Copilot/Codex CLI)
Models: OpenAI, Google Gemini, Ollama, vLLM, OptiLLM

Also in Coding agents and assistants

See all 63

Also in Coding agents and assistants
Rank	Project	Score
1	claw-codeRust CLI coding agent harness using Anthropic and OpenAI-compatible APIs194.9k starsMIT	83 out of 100
2	ArchonYAML workflow engine that runs AI coding agents through fixed dev steps23.7k starsMIT	80 out of 100
3	opencodeTerminal coding agent with build and plan modes, plus a desktop app213.5k starsMIT	79 out of 100
#4	piTerminal coding agent with extensions, skills, and an SDK115.2k starsMIT	79 out of 100
#5	oh-my-openagentTerminal coding agent with multi-model orchestration and git-backed memory70k starscustom license	79 out of 100

How to choose

Check which models it can use: your own keys, local models, or only the vendor's.
See what it may do without asking (edit files, run shell, push) and how you approve.
Prefer agents that work in a branch or worktree you can throw away.

Health

Releases: Not checked
Top contributor: Not checked
Security advisories: Not checked
Package advisories: Not checked
OpenSSF Scorecard: Not checked

Not read for this project yet: each refresh reads the facts of a few projects. Bots and coding-agent accounts are left out of the commit counts. Not part of the score: how we check.

License in plain words

Apache-2.0 Permissive

Use it at work: Yes
Change it and rebrand it: Yes
Sell it or host it for others: Yes

Your part: Keep the notices and mark the files you change. It also grants you the contributors’ patent rights.

A plain summary, not legal advice. Read the license on GitHub ↗ (opens in a new tab)

#52 of 63 in Coding agents and assistants on Archestack, as of 15 Oct 2026.

In your README

The badge updates twice a day.

[![Archestack](https://archestack.com/badge/algorithmicsuperintelligence/openevolve.svg)](https://archestack.com/best-of-vibe-coding/projects/openevolve/)

How badges work

On your site or blog

A card with the rank, score and signals, in light or dark to match the reader.

<iframe src="https://archestack.com/embed/best-of-vibe-coding/openevolve/" title="openevolve on Archestack" width="400" height="270" style="border:0;max-width:100%" loading="lazy"></iframe>

Preview the card

Best of Vibe Coding on GitHub

Each review comes from the project's README and facts we check on GitHub. The fix form opens with openevolve filled in.