GEPA vs Vellum
GEPA
gepa-ai.github.ioVellum
vellum.aiSide by side
| Figure | GEPA | Vellum |
|---|---|---|
| Price from | Free | $30/mo |
| Free plan | Yes | Yes |
| Free trial | — | — |
| Open source | — | Yes |
| Web | — | Yes |
| Windows | — | Yes |
| macOS | — | Yes |
| iOS | — | Yes |
| Android | — | Yes |
| API | Yes | Yes |
| Self-hosted | Yes | Yes |
| Plans published | 1 | 9 |
| Spec score · AI Prompt Generators | 8.6 (#4) | 8.6 (#5) |
| Integrations | The project lists integrations with DSPy, MLflow, Comet ML Opik, Pydantic AI, OpenAI Cookbook, Hugging Face Cookbook, and Google ADK. | OAuth integrations include GitHub, Notion, Linear, Airtable, Spotify, Gmail, Google Calendar, Slack, and other services, while MCP servers provide additional tools. |
| Support | The project points users to its documentation, GitHub issues, and a Discord community. | Vellum lists support contact as [email protected]. |
| What it does | GEPA is an LLM-based text evolution engine for optimizing prompts, code, configurations, agent architectures, policies, and other text-representable artifacts. | Vellum is a personal AI assistant that can read files, manage calendars, control computers, take calls, perform transactions, and retain personal context. |
Plans
Vellum
- BaseFree
- FreeFree
- Mighty$30 a month
- Mighty$30 a month
- Custom$50 a month
- Super$100 a month
- Super$100 a month
- Ultra$200 a month
- Ultra$200 a month
In their own words
GEPA
GEPA is a free, open-source Python package for optimizing prompts, code, configurations, agent architectures, policies, and other text-based artifacts. Its optimize_anything API connects to any system through an evaluator you write, so DSPy or another framework is not required. GEPA examines execution traces and evaluator feedback, uses language-model reflection to diagnose problems and propose targeted mutations, then selects candidates with Pareto-aware criteria. It can optimize API-only GPT, Claude, and Gemini models without model-weight access. Human-readable traces explain changes and support agent debugging. Built-in adapters cover Default, Confidence, DSPy Full Program, Generic RAG, MCP, TerminalBench, AnyMaths, and LangChain, with integrations including MLflow, Comet ML Opik, Pydantic AI, and cookbook projects. You can limit metric calls and set several stopping conditions. GEPA works through API or self-hosted deployment, installs with pip, and uses the MIT license.
Vellum
Vellum is a personal AI assistant that can read files, manage calendars, send email, run code, and control a computer. It saves preferences and context in its workspace for later conversations, and users can create, install, and manage skills. Listed OAuth services include Discord, GitHub, Google services, HubSpot, Linear, Twitter (X), Asana, Notion, Microsoft Outlook, and Todoist. Vellum says OAuth tokens stay in a credential vault rather than being exposed to the language model. Sensitive actions prompt for approval, and users can set trust rules for standing permissions. By default, conversation context is sent to Anthropic to generate responses; OpenAI, Google, OpenRouter, Fireworks, and local Ollama are listed alternatives. An internet connection is required. Vellum Cloud runs assistants in isolated environments with private storage and 24/7 availability. The Free plan includes first tasks and pay-as-you-go credits. Mighty costs $30 a month, Super $100 a month, and Ultra $200 a month. There is no shipped Linux desktop client, but Linux users can use the web app, Chrome extension, or self-host the runtime.