GEPAzilla
GEPAzilla is an open-source GEPA prompt optimizer for iterating on system prompts with datasets and scorers.
What GEPAzilla is
GEPAzilla is an open-source GEPA prompt optimizer for iterating on system prompts with datasets and scorers. You can load a dataset or use its sample, then combine deterministic checks and LLM judges with configurable scorer weights and duplicates. Validation rows remain outside reflection and rank candidate prompts on a Pareto frontier. Results show candidate prompts, iteration trends, scorer failure rates, averages, common notes, and a latency metric; request latency is optimized internally. GEPAzilla runs as a web app, API, or self-hosted Next.js project that can be built and started with pnpm. GEPA runs, scorer previews, and model discovery require a Vercel AI Gateway API key. The project says telemetry stays on your machine, including tokens, routing, costs, and errors, and that run data is not exfiltrated. Dataset tools currently accept JSON copied or pasted from the clipboard. It is released under the MIT License.
Good
- Open-source MIT-licensed project
- Supports deterministic and LLM-based scoring
- Holds validation rows out of reflection
- Includes latency and scorer diagnostics
- Available as web, API, or self-hosted
Watch out
- Requires a Vercel AI Gateway API key
- Dataset import currently uses clipboard JSON
- Users must provide their own model access
Who it suits
It suits curious builders experimenting with prompt optimization using their own API key. Self-hosting and API availability also fit users wanting deployment flexibility.
Verdict
GEPAzilla provides an open prompt-optimization workflow with held-out validation, diagnostics, and local telemetry. Its API-key requirement and clipboard-based dataset input shape the setup.
The facts
Each one read from GEPAzilla's own pages, with the date it was checked.
What it does
- API key requirement
- The project README says an AI Gateway API key is required for GEPA runs, scorer previews, and model discovery.checked 10 Oct 2026
- Built-in latency metric
- GEPA optimizes request latency internally, and the Results tab always includes a latency metric.checked 10 Oct 2026
- Data handling
- The README says telemetry stays in the browser, telemetry events are not sent to a remote service, and run data is not exfiltrated.checked 10 Oct 2026
- Data import limit
- Dataset tools currently support copying or pasting JSON payloads from the clipboard.checked 4 Oct 2026
- Dataset format limit
- Dataset tools currently support copying or pasting JSON payloads from the clipboard.checked 10 Oct 2026
- Dataset input
- Users can load their own rows or start with the sample dataset.checked 10 Oct 2026
- Deployment
- The README describes the application as a standard Next.js project that can be built and started locally.checked 10 Oct 2026
- Diagnostics
- Results include scorer failure rates, averages, and common notes, alongside candidate prompts and iteration trends.checked 4 Oct 2026
- Gateway integration
- The README specifies a Vercel AI Gateway credential for the required API key.checked 10 Oct 2026
- Intended users
- The maker describes the console as a way for curious builders to experiment with prompt optimization using their own API key.checked 4 Oct 2026
- License
- The repository states that GEPAzilla is released under the MIT License.checked 10 Oct 2026
- Model access
- The project requires users to supply an AI Gateway API key for GEPA runs, scorer previews, and model discovery; its README specifies Vercel AI Gateway.checked 4 Oct 2026
- Purpose
- GEPAzilla is an open-source GEPA prompt optimizer that uses datasets and custom metrics to find stronger prompts.checked 10 Oct 2026
- Scoring
- It supports combining deterministic checks with LLM judges and weighting or duplicating scorers.checked 10 Oct 2026
- Telemetry
- The maker says telemetry stays local and includes tokens, routing, costs, and errors for inspection.checked 10 Oct 2026
- Validation
- Validation rows are kept out of reflection and used to rank candidate prompts on the Pareto frontier.checked 4 Oct 2026
- Validation data
- Validation rows are kept untouched during reflection and used to rank candidates on the Pareto frontier.checked 10 Oct 2026
Alternatives to GEPAzilla
- 1
Vellum
Free planOpen sourceAndroidiOSWebAPI8.7From $30 a month - 2
Agenta
Free planOpen sourceWebAPISelf-hosted8.0From $29 a month - 3
Latitude
Free planOpen sourceWebAPILinuxSelf-hosted7.7From $99 a month - 4
VantagePrompt
Free planWeb7.3From $3 a month - 5
PromptLayer
Free planWebAPISelf-hosted7.0From $49 a month - 6
PromptWizard
Free planAPILinuxmacOSSelf-hosted6.8Free plan