Skip to main content

​
TL;DR

  • In-Process Task Apps: Complete MIPRO and GEPA optimization with automatic task app and tunnel management
  • Gemini 3 Model Support: Google’s latest Gemini 3 models now supported
  • Production-Ready Examples: Complete pipeline examples for GEPA and MIPRO
  • SDK Release 0.2.25.dev1: Latest package updates

​
In-Process Task Apps for GEPA & MIPRO

Complete in-process optimization workflows with automatic task app and tunnel management.

​
MIPRO In-Process Demo

  • Single-Script Execution: Everything runs in one Python script - no separate terminals or manual process management
  • Banking77 Demo: Intent classification task with 77 banking categories using PolyAI/banking77 dataset
  • Automatic Cleanup: Task app server and Cloudflare tunnel cleaned up automatically on exit
  • Command: uvx synth-ai demo mipro scaffolds complete demo to demo_mipro/ directory

​
GEPA In-Process Demo

  • Heart Disease Classification: Medical classification task using buio/heart-disease dataset
  • Evolutionary Optimization: Population-based optimization with mutation, crossover, and Pareto selection
  • Multi-Objective: Balances accuracy, token efficiency, and tool-calling behavior
  • Example Script: examples/blog_posts/gepa/run_fully_in_process.py demonstrates complete workflow

​
Gemini 3 Model Support

Google Gemini 3 models are now supported across prompt optimization workflows:
  • Supported Models: gemini-3-pro, gemini-3-flash, gemini-3-flash-lite for policy and meta-models
  • Provider Integration: Full support for provider = "google" in prompt optimization configurations
  • Cost Tracking: Automatic pricing calculation and token tracking for Gemini 3 models
  • Example Configs: Gemini 3 configurations available in MIPRO and GEPA example configs

​
Cloudflare Tunnel Enhancements

Improved Cloudflare tunnel management for prompt optimization workflows:
  • Automatic Tunnel Discovery: InProcessTaskApp automatically creates and manages tunnels
  • Health Check Integration: Waits for task app health before opening tunnel
  • Process Management: Background thread management with graceful shutdown
  • Cross-Platform Support: Works on macOS, Linux, and Windows with automatic cloudflared installation

​
Production-Ready Examples

New production-ready examples demonstrating prompt optimization workflows:

​
Complete Pipeline Examples

  • GEPA Pipeline: run_gepa_example.py - Complete GEPA optimization workflow from baseline evaluation to final prompt deployment
  • MIPRO Pipeline: run_mipro_example.py - Complete MIPRO optimization workflow with programmatic polling and progress tracking
  • In-Process Task Apps: Automatic task app management with Cloudflare tunnel support for production deployments
  • Self-Contained Scripts: Everything in one script - no external dependencies or manual setup required

​
Production Integration Features

  • In-Process Task App Management: InProcessTaskApp utility automatically manages FastAPI servers and Cloudflare tunnels
  • Programmatic Polling: Built-in job status polling with progress callbacks and timeout handling
  • Prompt Retrieval: Easy extraction of optimized prompts from completed jobs
  • Baseline & Final Evaluation: Complete evaluation pipeline comparing initial vs optimized prompts

​
SDK Release 0.2.25.dev1

  • Version Bump: Updated synth-ai package to 0.2.25.dev1
  • Modal Deployment: Updated default SYNTH_AI_VERSION in Modal deployments to 0.2.25.dev1

​
Documentation

  • MIPRO Quickstart: Complete quickstart guide at /quickstart/prompt-optimization-mipro
  • GEPA In-Process Guide: Explanatory guide at /quickstart/gepa-in-process
  • Production Guide: Comprehensive guide at /blog/prompt-optimization-benchmarks
  • Examples Directory: Complete examples in examples/blog_posts/vendored_prompt_learning/

​
Use Cases

  • Rapid Prototyping: Test prompt optimization without infrastructure setup
  • Local Development: Develop and debug optimization workflows entirely locally
  • A/B Testing: Automatically find better prompts for your use case without manual intervention
  • Performance Tuning: Continuously improve prompt performance as your data changes