gepa
A framework for optimizing textual system components (AI prompts, code snippets, etc.) using LLM-based reflection and Pareto-efficient evolutionary search.
Why this rank:Strong adoptionRelease freshnessHealthy release cadence
Description
<p align="center"> <img src="https://raw.githubusercontent.com/gepa-ai/gepa/refs/heads/main/assets/gepa_logo_with_text.svg" alt="GEPA Logo" width="450"> </p> <p align="center"> <strong>Optimize any text parameter â prompts, code, agent architectures, configurations â using LLM-based reflection and Pareto-efficient evolutionary search.</strong> </p> <p align="center"> <a href="https://gepa-ai.github.io/gepa/"><strong>Website</strong></a>  |  <a href="https://gepa-ai.github.io/gepa/guides/quickstart/"><strong>Quick Start</strong></a>  |  <a href="https://arxiv.org/abs/2507.19457"><strong>Paper</strong></a>  |  <a href="https://gepa-ai.github.io/gepa/blog/"><strong>Blog</strong></a>  |  <a href="https://discord.gg/WXFSeVGdbW"><strong>Discord</strong></a> </p> <p align="center"> <a href="https://pypi.org/project/gepa/"><img src="https://img.shields.io/pypi/v/gepa?logo=python&logoColor=white&color=3776ab" alt="PyPI"></a> <a href="https://pepy.tech/projects/gepa"><img src="https://static.pepy.tech/badge/gepa" alt="Downloads"></a> <a href="https://github.com/gepa-ai/gepa"><img src="https://img.shields.io/github/stars/gepa-ai/gepa?style=flat&logo=github&color=181717" alt="GitHub stars"></a> <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-green?style=flat" alt="License"></a> </p> <p align="center"> <a href="https://join.slack.com/t/gepa-ai/shared_invite/zt-3o352xhyf-QZDfwmMpiQjsvoSYo7M1_w"><img src="https://badgen.net/badge/icon/Slack?icon=slack&label&color=4A154B" alt="Slack"></a> <a href="https://discord.gg/WXFSeVGdbW"><img src="https://dcbadge.limes.pink/api/server/https://discord.gg/WXFSeVGdbW?style=flat" alt="Discord"></a> </p> --- ## What is GEPA? **GEPA** (Genetic-Pareto) is a framework for optimizing any system with textual parameters against any evaluation metric. Unlike RL or gradient-based methods that collapse execution traces into a single scalar reward, GEPA uses LLMs to *read* full execution traces â error messages, profiling data, reasoning logs â to diagnose *why* a candidate failed and propose targeted fixes. Through iterative reflection, mutation, and Pareto-aware selection, GEPA evolves high-performing variants with minimal evaluations. **If you can measure it, you can optimize it**: prompts, code, agent architectures, scheduling policies, vector graphics, and more. ### Key Results | | | |---|---| | **90x cheaper** | Open-source models + GEPA beat Claude Opus 4.1 at [Databricks](https://www.databricks.com/blog/building-state-art-enterprise-agents-90x-cheaper-automated-prompt-optimization) | | **35x faster than RL** | 100â500 evaluations vs. 5,000â25,000+ for GRPO ([paper](https://arxiv.org/abs/2507.19457)) | | **32% â 89%** | ARC-AGI agent accuracy via [architecture discovery](https://gepa-ai.github.io/gepa/blog/introducing-optimize-anything/#5-agent-architecture-discovery) | | **40.2% cost savings** | Cloud scheduling policy [discovered by GEPA](https://gepa-ai.github.io/gepa/blog/introducing-optimize-anything/#3-systems-research), beating expert heuristics | | **55% â 82%** | Coding agent resolve rate on Jinja via [auto-learned skills](https://gepa-ai.github.io/gepa/blog/automatically-learning-skills-for-coding-agents/) | | **50+ production uses** | Across Shopify, Databricks, Dropbox, OpenAI, Pydantic, MLflow, Comet ML, and [more](https://gepa-ai.github.io/gepa/guides/use-cases/) | > *"Both DSPy and (especially) **GEPA are currently severely under hyped** in the AI context engineering world"* â **Tobi Lutke**, CEO, Shopify --- ## Installation ```bash pip install gepa ``` To install the latest from `main`: ```bash pip install git+https://github.com/gepa-ai/gepa.git ``` --- ## Quick Start ### Simple Prompt Optimization Optimize a system prompt for math problems from the AIME benchmark in a few lines of code ([full tutorial](https://dspy.ai/tutorials/gepa_aime/)): ```python import gepa trainset, valset, _ = gepa.examples.aime.init_dataset() seed_prompt = { "system_prompt": "You are a helpful assistant. Answer the question. " "Put your final answer in the format '### <answer>'" } result = gepa.optimize( seed_candidate=seed_prompt, trainset=trainset, valset=valset, task_lm="openai/gpt-4.1-mini", max_metric_calls=150, reflection_lm="openai/gpt-5", ) print("Optimized prompt:", result.best_candidate['system_prompt']) ``` **Result:** GPT-4.1 Mini goes from 46.6% â 56.6% on AIME 2025 (+10 percentage points). ### With DSPy (Recommended for AI Pipelines) The most powerful way to use GEPA for prompt optimization is within [DSPy](https://dspy.ai/), where it's available as `dspy.GEPA`. See [dspy.GEPA tutorials](https://dspy.ai/tutorials/gepa_ai_program/) for executable notebooks. ```python import dspy optimizer = dspy.GEPA( metric=your_metric, max_metric_calls=150, reflection_lm="openai/gpt-5", ) optimized
Release History
| Version | Changes | Urgency | Date |
|---|---|---|---|
| 0.1.1 | Imported from PyPI (0.1.1) | Low | 4/21/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.1 | ### Highlights **Candidate Tree Visualization** â GEPA now generates an interactive HTML lineage tree of all candidates explored during optimization. Nodes are color-coded by role (best, Pareto front, seed) with hover previews and click-to-pin tooltips for reading full prompt text. The tree is automatically logged to WandB and MLflow at each step â no configuration needed. ([#256](https://github.com/gepa-ai/gepa/pull/256)) **gskill: Automated Skill Learning for Coding Agents** â A new `gep | Low | 3/16/2026 |
| v0.1.0 | Introducing optimize_anything, GEPA's new API for optimizing any text parameter. Checkout the blogpost at https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-optimize-anything/ ## What's Changed * Automated docs and website by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/176 * Make ARC_AGI example readable in docs by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/177 * Add GA by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/179 * Add more external r | Low | 2/19/2026 |
| v0.0.27 | ## What's Changed * Update README by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/174 * Add callback system for optimization instrumentation by @dbreunig in https://github.com/gepa-ai/gepa/pull/155 ## New Contributors * @dbreunig made their first contribution in https://github.com/gepa-ai/gepa/pull/155 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.26...v0.0.27 | Low | 1/28/2026 |
| v0.0.26 | ## What's Changed * Add cache by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/166 * Fix a 1-line bug where cloudpickle wasn't used sometimes when it should by @4gatepylon in https://github.com/gepa-ai/gepa/pull/167 * Validate seed_candidate at optimize() entry point by @Copilot in https://github.com/gepa-ai/gepa/pull/169 * Fix MLflow logging, and add tracking for how many metric calls have been issued by @chenmoneygithub in https://github.com/gepa-ai/gepa/pull/171 ## New Contr | Low | 1/24/2026 |
| v0.0.25 | ## What's Changed * Add Python 3.14 support by @Copilot in https://github.com/gepa-ai/gepa/pull/165 ## New Contributors * @Copilot made their first contribution in https://github.com/gepa-ai/gepa/pull/165 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.24...v0.0.25 | Low | 1/20/2026 |
| v0.0.24 | ## What's Changed * Allow for objective and hybrid pareto frontier tracking by @MatsErdkamp in https://github.com/gepa-ai/gepa/pull/92 ## New Contributors * @MatsErdkamp made their first contribution in https://github.com/gepa-ai/gepa/pull/92 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.23...v0.0.24 | Low | 1/5/2026 |
| v0.0.23 | ## What's Changed * Update README.md by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/135 * Add HuggingFace Cookbook link for prompt optimization by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/136 * Fix/145 write outputs not scores by @05satyam in https://github.com/gepa-ai/gepa/pull/151 * feat(default_adapter): configurable eval returns score + feedback (#62) by @villurignanesh in https://github.com/gepa-ai/gepa/pull/147 * Add MCP (Model Context Protocol) Adapter Su | Low | 12/28/2025 |
| v0.0.22 | ## What's Changed * Skip components that are not present in reflective dataset instead of raising KeyError failure by @heyalexchoi in https://github.com/gepa-ai/gepa/pull/76 ## New Contributors * @heyalexchoi made their first contribution in https://github.com/gepa-ai/gepa/pull/76 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.21...v0.0.22 | Low | 11/10/2025 |
| v0.0.21 | ## What's Changed * Add logging option for individual valset scores by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/131 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.20...v0.0.21 | Low | 11/8/2025 |
| v0.0.20 | ## What's Changed * Enhance types by @mwildehahn in https://github.com/gepa-ai/gepa/pull/93 * Update README with new GEPA projects and articles by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/129 * Remove 'stubPath' from pyrightconfig.json by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/130 ## New Contributors * @mwildehahn made their first contribution in https://github.com/gepa-ai/gepa/pull/93 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.19 | Low | 11/7/2025 |
| v0.0.19 | ## What's Changed * Dynamic Validation in GEPAState by @aria42 in https://github.com/gepa-ai/gepa/pull/100 * Allowing custom batch_sampler by @CrisRodriguez in https://github.com/gepa-ai/gepa/pull/119 * epsilon greedy candidate selector by @vacmar01 in https://github.com/gepa-ai/gepa/pull/109 * Add batch_completion_kwargs by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/127 ## New Contributors * @aria42 made their first contribution in https://github.com/gepa-ai/gepa/pull/100 | Low | 11/4/2025 |
| v0.0.18 | ## What's Changed * Mock wandb and mlflow instead of testing storage by @TomeHirata in https://github.com/gepa-ai/gepa/pull/89 * Update README.md by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/90 * Fix parsing for ``` blocks with language in InstructionProposalSignature by @CDBiddulph in https://github.com/gepa-ai/gepa/pull/99 * Fix reflective proposer stalling by consuming metric budget on trace failures by @fgn in https://github.com/gepa-ai/gepa/pull/98 * Highlight GEPA Useca | Low | 10/25/2025 |
| v0.0.17 | ## What's Changed * Add reference to ATLAS+GEPA for incident diagnosis by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/84 * Add new GEPA-related resources to README by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/88 * fix: gepa state initialization from run_dir by @enitrat in https://github.com/gepa-ai/gepa/pull/86 ## New Contributors * @enitrat made their first contribution in https://github.com/gepa-ai/gepa/pull/86 **Full Changelog**: https://github.com/gepa-a | Low | 9/25/2025 |
| v0.0.16 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.15...v0.0.16 | Low | 9/24/2025 |
| v0.0.15-alpha.1 | ## What's Changed * Add link for 100% accuracy using GEPA on clock-hands problem by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/78 * Update README.md by @LakshyAAAgrawalP in https://github.com/gepa-ai/gepa/pull/82 * Bug Fix: copy base valset eval output as pareto front by @aobo-y in https://github.com/gepa-ai/gepa/pull/81 * Improve logging for subsample scores in GEPA engine by @CDBiddulph in https://github.com/gepa-ai/gepa/pull/83 ## New Contributors * @LakshyAAAgrawalP mad | Low | 9/24/2025 |
| v0.0.15 | ## What's Changed * Add link for 100% accuracy using GEPA on clock-hands problem by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/78 * Update README.md by @LakshyAAAgrawalP in https://github.com/gepa-ai/gepa/pull/82 * Bug Fix: copy base valset eval output as pareto front by @aobo-y in https://github.com/gepa-ai/gepa/pull/81 * Improve logging for subsample scores in GEPA engine by @CDBiddulph in https://github.com/gepa-ai/gepa/pull/83 ## New Contributors * @LakshyAAAgrawalP mad | Low | 9/24/2025 |
| v0.0.14 | ## What's Changed * Add max candidates stopper by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/77 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.13...v0.0.14 | Low | 9/22/2025 |
| v0.0.13 | ## What's Changed * Add link for multi-agent system in healthcare using DSPy by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/68 * RAG Optimization: Add Generic RAG Adapter with support for major vector databases by @Shashikant86 in https://github.com/gepa-ai/gepa/pull/60 * Add GEPA Talk Slides link to README by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/73 * Safely stop run by @ShriyaaNarayanan in https://github.com/gepa-ai/gepa/pull/64 ## New Contributors * @Sh | Low | 9/22/2025 |
| v0.0.12 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.11...v0.0.12 | Low | 9/9/2025 |
| v0.0.11 | ## What's Changed * Add video link to README by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/63 * Add 'all' component selector for selecting all components by @andressrg in https://github.com/gepa-ai/gepa/pull/67 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.10...v0.0.11 | Low | 9/9/2025 |
| v0.0.10 | ## What's Changed * Introduce mlflow by @TomeHirata in https://github.com/gepa-ai/gepa/pull/51 ## New Contributors * @TomeHirata made their first contribution in https://github.com/gepa-ai/gepa/pull/51 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.9...v0.0.10 | Low | 9/5/2025 |
| v0.0.9 | ## What's Changed * Added `google-auth` as dependency to make LiteLLM Google Vertex-friendly ... by @egmaminta in https://github.com/gepa-ai/gepa/pull/41 * Revise README with additional GEPA resources and clarity by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/55 * Add ARC-AGI example by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/57 * Add custom component selector support to optimize() by @andressrg in https://github.com/gepa-ai/gepa/pull/52 ## New Contributors | Low | 9/5/2025 |
| v0.0.8 | ## What's Changed * feat: improved the README.md for AnyMaths; fixed AnyMathsAdapter code; provided example training and dataset preparation; added eval results and sample eval code by @egmaminta in https://github.com/gepa-ai/gepa/pull/15 * Enhance README with Discord badge and video tutorial by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/37 * GEPA to evolve the full DSPy program, including structure by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/38 * Fix long outpu | Low | 9/3/2025 |
| v0.0.7 | ## What's Changed * Fix TQDM by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/32 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.6...v0.0.7 | Low | 8/25/2025 |
| v0.0.6 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5...v0.0.6 | Low | 8/25/2025 |
| v0.0.6-alpha.3 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.6-alpha.2...v0.0.6-alpha.3 | Low | 8/25/2025 |
| v0.0.6-alpha.2 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.6-alpha.1...v0.0.6-alpha.2 | Low | 8/25/2025 |
| v0.0.6-alpha.1 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5...v0.0.6-alpha.1 | Low | 8/25/2025 |
| v0.0.5 | ## What's Changed * feat: GEPAAdapter for any maths, reasoning dataset in HuggingFace; Ollama-friendly by @egmaminta in https://github.com/gepa-ai/gepa/pull/7 * feat: add terminal bench example by @Shangyint in https://github.com/gepa-ai/gepa/pull/6 * misc: split tb adapter into adapter folder by @Shangyint in https://github.com/gepa-ai/gepa/pull/8 * fix: add more experiments and fix path issues by @Shangyint in https://github.com/gepa-ai/gepa/pull/11 * Add Replayable AIME Prompt Optimizati | Low | 8/24/2025 |
| v0.0.5-alpha.13 | **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.12...v0.0.5-alpha.13 | Low | 8/24/2025 |
| v0.0.5-alpha.12 | ## What's Changed * Fix wait-on-check by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/23 * Fix names by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/24 * Remove status checks by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/25 * Remove check status by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/26 * Fix uv venv by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/27 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5 | Low | 8/24/2025 |
| v0.0.5-alpha.11 | ## What's Changed * Fix uv venv by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/27 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.10...v0.0.5-alpha.11 | Low | 8/24/2025 |
| v0.0.5-alpha.10 | ## What's Changed * Remove check status by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/26 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.9...v0.0.5-alpha.10 | Low | 8/24/2025 |
| v0.0.5-alpha.9 | ## What's Changed * Remove status checks by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/25 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.8...v0.0.5-alpha.9 | Low | 8/24/2025 |
| v0.0.5-alpha.8 | ## What's Changed * Fix names by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/24 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.7...v0.0.5-alpha.8 | Low | 8/24/2025 |
| v0.0.5-alpha.7 | ## What's Changed * Fix wait-on-check by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/23 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.6...v0.0.5-alpha.7 | Low | 8/24/2025 |
| v0.0.5-alpha.6 | ## What's Changed * Try running tests on tag pushes by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/22 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.5...v0.0.5-alpha.6 | Low | 8/24/2025 |
| v0.0.5-alpha.5 | ## What's Changed * Fix workflow by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/21 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.4...v0.0.5-alpha.5 | Low | 8/24/2025 |
| v0.0.5-alpha.4 | ## What's Changed * Fix status checks by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/20 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.3...v0.0.5-alpha.4 | Low | 8/24/2025 |
| v0.0.5-alpha.3 | ## What's Changed * Fix circular checks by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/19 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.2...v0.0.5-alpha.3 | Low | 8/24/2025 |
| v0.0.5-alpha.2 | ## What's Changed * Fix token by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/18 **Full Changelog**: https://github.com/gepa-ai/gepa/compare/v0.0.5-alpha.1...v0.0.5-alpha.2 | Low | 8/24/2025 |
| v0.0.5-alpha.1 | ## What's Changed * Make v0.0.1 of GEPA by @LakshyAAAgrawal in https://github.com/gepa-ai/gepa/pull/1 * feat: GEPAAdapter for any maths, reasoning dataset in HuggingFace; Ollama-friendly by @egmaminta in https://github.com/gepa-ai/gepa/pull/7 * feat: add terminal bench example by @Shangyint in https://github.com/gepa-ai/gepa/pull/6 * misc: split tb adapter into adapter folder by @Shangyint in https://github.com/gepa-ai/gepa/pull/8 * fix: add more experiments and fix path issues by @Shangyin | Low | 8/23/2025 |
Dependencies & License Audit
Loading dependencies...
Similar Packages
More from pypi
markitdownUtility tool for converting various files to Markdown
fastapiFastAPI framework, high performance, easy to learn, fast to code, ready for production
djangoA high-level Python web framework that encourages rapid development and clean, pragmatic design.
flaskA simple framework for building complex web applications.
More in Frameworks
langchainThe agent engineering platform
deer-flowAn open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of ta
tqdmFast, Extensible Progress Meter
simBuild, deploy, and orchestrate AI agents. Sim is the central intelligence layer for your AI workforce.