Skip to main content
Graph Evolve is Synth AI’s product surface for building reliable LLM systems. You bring a dataset of real inputs and what “good” looks like. Graph Evolve learns an optimized prompt graph and gives you a stable way to serve it. Unlike our core prompt-optimization APIs, Graph Evolve is designed to feel like software you ship, not machinery you configure.

​
What you get

  • Dataset‑in → graph‑out: upload a GraphEvolveTaskSet and we train a prompt graph end‑to‑end.
  • Built‑in verifying: rubric, contrastive, or gold‑examples scoring without writing verifier code.
  • Live progress: poll status or stream events/metrics while training runs.
  • Downloadable artifacts: fetch the best prompt snapshot for local use.
  • Production inference: call /api/graphgen/graph/completions to run the optimized graph on new JSON inputs.
  • Massive context support: use graph_type: "rlm" for 1M+ token context via tool-based search (see RLM graphs).
Today Graph Evolve uses GEPA under the hood because it’s the fastest way to improve graphs from data. We’ll add additional improvement paths over time while keeping the same dataset and API surface.

​
How it works

  1. Define your task as data: tasks are your real inputs; gold outputs and/or rubrics define success.
  2. Train: POST /api/graphgen/jobs starts a Graph Evolve training run.
  3. Monitor: GET /api/graphgen/jobs/{graph_gen_id} or stream events.
  4. Use: download the best prompt or serve it via /api/graphgen/graph/completions.
Get started here: Graphs quickstart.

​
SDK + cookbooks

  • Python SDK: see Graph jobs for the GraphOptimizationJob API.
  • Examples: see Graphs cookbooks for style matching and generative graphs.

​
Pricing

Graph Evolve training and inference are usage‑based. See Pricing overview.