Playground

Try Hanzo in your browser

Hands-on sandbox for every Hanzo product. No install, no signup for read-only flows. Run models, prompt agents, wire up MCP.

Pick what to try

Each card is a working sandbox.

Zen Models

Run prompts against 390+ open-weight models. Compare side-by-side.

Try a model

Bot

Spin up a multi-agent simulation in the browser. No setup.

Launch a bot

MCP

Connect tools to Claude/Cursor/Hanzo Dev with one click.

Browse MCP servers

Hanzo Dev

AI engineer that ships PRs from a sentence.

Run Dev

Chat

Talk to 14 Zen models + 100+ third-party LLMs with MCP tools.

Open Chat

Agents

Multi-agent SDK with tool use, planning, memory.

Read the SDK

Or copy-paste a starter

Three ways in. All open source.

curlHello Zen
curl https://api.hanzo.ai/v1/chat/completions \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -d '{"model":"zen-2","messages":[{"role":"user","content":"Hello"}]}'
pythonAgents SDK
from hanzo import Agent
agent = Agent(model="zen-2", tools=["search", "browser"])
print(agent.run("Find the latest AI papers and summarize"))
typescriptMCP client
import { Client } from '@hanzo/mcp'
const c = new Client({ server: 'hanzo://search' })
const r = await c.call('search', { query: 'rust async runtime' })

Open source DX, end to end

Playground · Dev (CLI agent) · MCP · Hanzo CLI · Extensions · SDKs in 4 languages. All on GitHub.

Open source

License: Apache-2.0hanzoai

Get Playground

Hands-on sandbox for every product