# Zen Models by Zoo Labs Foundation — Hanzo AI

> Zen from Zoo Labs Foundation: frontier AI models from 0.6B to 1T+ parameters. Zen MoDE (Mixture of Diverse Experts). OpenAI-compatible API. Vision, audio, code, math. From $0.15/MTok.

45+ Zen Models

# Zen ModelsLanguage · Code · Vision · Audio

45+ open-weight models from 0.6B to 1T+, built by [Zoo Labs Foundation](https://zoo.industries) and served on the Hanzo API. Zen MoDE (Mixture of Diverse Experts) architecture. All models on HuggingFace.

[Explore All Models](https://hanzo.ai/zen/models)[View Pricing](https://hanzo.ai/pricing)

[Get API Key](https://console.hanzo.ai)[HuggingFace](https://huggingface.co/zenlm)[GitHub](https://github.com/zenlm/zen)

zen-models

45+

Models

1.04T

Max Params

2M

Max Context

—

From

Language, code, vision, audio, video

Zen MoDE — Mixture of Diverse Experts

One endpoint, one key

Weights published for most of them

## Full Model Catalog

45+ models · Available via API and HuggingFace

[Browse on HuggingFace](https://huggingface.co/zenlm)

EdgeProMaxUltraVisionMultimodal

[zen5Pro35B3B active1M ctx1m context](https://huggingface.co/zenlm/zen5-gguf)[zen5-miniProTBA10B active1M ctx1m context](https://huggingface.co/zenlm/zen5-mini-gguf)[zen5-coderPro80B3B active1M ctx1m context](https://huggingface.co/zenlm/zen5-coder-gguf)[zen5-proProTBA1M ctx1m context](https://huggingface.co/zenlm/zen5-pro-gguf)[zen5-maxProTBA1M ctx1m context](https://huggingface.co/zenlm/zen5-max-gguf)[zen3-omniPro30B3B active131K ctx131k context](https://huggingface.co/zenlm/zen3-omni)[zen3-vlPro8B128K ctx128k context](https://huggingface.co/zenlm/zen3-vl)[zen3-guardPro5B128K ctx128k context](https://huggingface.co/zenlm/zen3-guard)[zen5-embeddingCloud APIProTBA33K ctx32k context](https://hanzo.ai/zen/models)[zen3-embeddingCloud APIProTBA33K ctx32k context](https://hanzo.ai/zen/models)[zen-rerankCloud APIProTBAcross-encoder rerank](https://hanzo.ai/zen/models)[zen-imageCloud APIVisionTBAtext-to-image](https://hanzo.ai/zen/models)[zen-voiceCloud APIProTBAtext-to-speech](https://hanzo.ai/zen/models)[zen-musicCloud APIProTBAtext-to-music](https://hanzo.ai/zen/models)[zen-foleyCloud APIProTBAtext-to-sound-effect](https://hanzo.ai/zen/models)[zen-videoCloud APIProTBAtext-to-video](https://hanzo.ai/zen/models)[zen3-nanoPro8B41K ctx40k context](https://huggingface.co/zenlm/zen3-nano)[zen-nanoPro0.6B41K ctx40k context](https://huggingface.co/zenlm/zen-nano-0.6b)[zen-ecoPro4B262K ctx262k context](https://huggingface.co/zenlm/zen-eco-4b-instruct)[zen-embedding-8bPro8B41K ctx40k context](https://huggingface.co/zenlm/zen-embedding-8B)[zen-embedding-4bPro4B41K ctx40k context](https://huggingface.co/zenlm/zen-embedding-4B)[zen-embedding-0.6bPro0.6B33K ctx32k context](https://huggingface.co/zenlm/zen-embedding-0.6B)[zen-reranker-8bPro8B41K ctx40k context](https://huggingface.co/zenlm/zen-reranker-8B)[zen-reranker-4bPro4B41K ctx40k context](https://huggingface.co/zenlm/zen-reranker-4B)[zen-reranker-0.6bPro0.6B41K ctx40k context](https://huggingface.co/zenlm/zen-reranker-0.6B)[zen3-imageVisionTBAtext-to-image](https://huggingface.co/zenlm/zen3-image)[zen3-ttsPro1.7B12 hz codec](https://huggingface.co/zenlm/zen3-tts)[zen3-asrPro1.7B30 languages](https://huggingface.co/zenlm/zen3-asr)[zen-musicianPro7Btext-to-music](https://huggingface.co/zenlm/zen-musician)[zen-agent-4bPro4B262K ctx262k context](https://huggingface.co/zenlm/zen-agent-4b)[zen-routerPro0.6B32K ctxencoder-pooled route](https://huggingface.co/zenlm/zen-router)[zen-medicalPro32B41K ctx40k context](https://huggingface.co/zenlm/zen-medical)[zen-legalPro32B41K ctx40k context](https://huggingface.co/zenlm/zen-legal)[zen-financePro32B41K ctx40k context](https://huggingface.co/zenlm/zen-finance)[zen-sqlPro32B41K ctx40k context](https://huggingface.co/zenlm/zen-sql)[zen-translatePro32B41K ctx40k context](https://huggingface.co/zenlm/zen-translate)[zen-multilingualPro32B41K ctx40k context](https://huggingface.co/zenlm/zen-multilingual)[zen-worldPro14Btext-to-video](https://huggingface.co/zenlm/zen-world)[zen-directorVision5Btext + image to vide](https://huggingface.co/zenlm/zen-director)[zen-image-editVisionTBAimage editing](https://huggingface.co/zenlm/zen-image-edit)[zen-3dVisionTBAimage-to-3d](https://huggingface.co/zenlm/zen-3d)[zen-dub-liveProTBAlive dubbing](https://huggingface.co/zenlm/zen-dub-live)[zen-designer-235b-a22b-instructPro235B22B active131K ctx131k context](https://huggingface.co/zenlm/zen-designer-235b-a22b-instruct)[zen-designer-235b-a22b-thinkingPro235B22B active131K ctx131k context](https://huggingface.co/zenlm/zen-designer-235b-a22b-thinking)[zen-designer-ggufPro235B22B active131K ctxgguf weights](https://huggingface.co/zenlm/zen-designer-gguf)

[View Full Specs & Pricing](https://hanzo.ai/zen/models)[Open Source on GitHub](https://github.com/zenlm/zen)

## What you get by using ours

### Same request shape

Change the model name in a call you already wrote. Nothing else about the request changes.

### Weights you can take

Most Zen models are published on HuggingFace, so you can serve them yourself. A few are API-only; the catalog above marks which.

### 0.6B to frontier

Small enough to run on a laptop, or large enough to reason. Text, code, vision, audio, image and video, one API.

### Sparse where it counts

The Mixture-of-Diverse-Experts models hold a large parameter count but activate a small slice of it per token, so capacity is not what you pay for.

## Pick by what you are doing

Different models, one endpoint. The name in the request is the only thing that changes

### Language

Chat, reasoning, retrieval

zen5

zen5-pro

zen5-max

### Code

A million tokens of it

zen5-coder

zen5-mini

zen-sql

### Vision

Read a screen, draw a thing

zen3-vl

zen3-omni

zen3-image

### Audio

Listen, speak, compose

zen3-asr

zen3-tts

zen-music

In Training

Next Generation

## Zen 5 Ultra

A 2T-parameter MoDE, training now inside NVIDIA TEE confidential compute on [hanzo.network](https://hanzo.network). The point of doing it that way is that the run attests to itself: what data went in and what came out is checkable afterwards rather than taken on our word.

2T+

Parameters

MoDE

Mixture of Diverse Experts

TEE

On-Chain Verifiable Training

[Zen 5 Research Preview](https://hanzo.ai/research-access)[On-Chain Training](https://hanzo.network)[Current Catalog](https://hanzo.ai/zen/models)

## Get Started

[### The catalog Every model, with its parameters, context window and what it is for](https://hanzo.ai/zen/models)

[### What it costs Price per million tokens, per model, with nothing rounded in our favour](https://hanzo.ai/pricing)

[### A key Sign in, take a key, send the first request](https://console.hanzo.ai)

䷀ ䷸ ䷹ ䷺ ䷻ ䷼ ䷽ ䷾ ䷿ ䷡

## Why they are named this way

These models are built on eight engineering principles drawn from the 64 hexagrams of the I-Ching — orthogonality, smallness, completeness, clarity and composability among them. Ancient pattern language for systems that last.

[易經 · Explore the Full Philosophy](https://hanzo.ai/philosophy)

## Call one and see

Start with zen5-mini because it is fast and cheap, and move up only where a harder question needs it.

[Explore All Models](https://hanzo.ai/zen/models)[Try Hanzo Dev](https://hanzo.ai/dev)[Get API Key](https://console.hanzo.ai)

## Read the specs, or read the code

[Read the docs](https://docs.hanzo.ai/docs/services/models)[View on GitHub](https://github.com/hanzoai)

## Open source

License: Apache-2.0[zenlm/zen](https://github.com/zenlm/zen)

## Get Zen Models

Open-weight frontier models

[Deploy to Cloud](https://console.hanzo.ai/deploy)[Self-host](https://docs.hanzo.ai/docs/services/models)