Capability · Framework — orchestration

Genkit

Genkit is Google's answer to LangChain: an opinionated framework for defining flows, tools, retrievers, and prompts that runs the same whether you deploy to Firebase, Cloud Run, or your own server. It ships with a local developer UI for tracing and evaluating flows and supports Gemini, Claude, Llama, and any OpenAI-compatible model out of the box.

Framework facts

Category: orchestration
Language: TypeScript / Go / Python
License: Apache-2.0
Repository: https://github.com/firebase/genkit

Install

npm install genkit @genkit-ai/googleai
# or
pip install genkit
# local dev UI:
npm install -g genkit-cli && genkit start

Quickstart

import { genkit } from 'genkit';
import { googleAI, gemini15Flash } from '@genkit-ai/googleai';

const ai = genkit({ plugins: [googleAI()] });

export const summarize = ai.defineFlow(
  { name: 'summarize', inputSchema: z.string(), outputSchema: z.string() },
  async (text) => {
    const { text: out } = await ai.generate({
      model: gemini15Flash,
      prompt: `Summarize: ${text}`,
    });
    return out;
  }
);

Alternatives

LangChain / LangGraph — larger ecosystem
Vercel AI SDK — similar JS-first ergonomics
LlamaIndex — more RAG-opinionated
Semantic Kernel — Microsoft's equivalent

Frequently asked questions

Is Genkit tied to Google Cloud?

No. The core is model- and cloud-agnostic, and the provider plugin system supports Gemini, Anthropic, Ollama, OpenAI, and more. Firebase and Vertex AI are just the first-class deployment targets.

Genkit vs Vercel AI SDK?

Vercel AI SDK is more UI-focused (streaming React components out of the box). Genkit puts more weight on flows, evals, and production tracing. Many teams combine them — Genkit on the server, AI SDK on the client.

Sources

Genkit — docs — accessed 2026-04-20
Genkit GitHub — accessed 2026-04-20