⚡ Zen LM
Models

zen-pro

Professional-grade 32B dense model with 19K tokens/sec throughput.

zen-pro

Professional

A 32B dense transformer for professional workloads requiring high-quality reasoning and generation. Achieves 19K tokens/sec -- the strongest dense model in the foundation tier before stepping up to MoE architectures.

Specifications

PropertyValue
Model IDzen-pro
Parameters32B
ArchitectureDense
Context Window32K tokens
Throughput19K tokens/sec
StatusAvailable
HuggingFacezenlm/zen-pro-32b

Capabilities

  • High-quality reasoning and analysis
  • Professional document drafting
  • Complex instruction following
  • Code generation and review
  • Multilingual translation and understanding
  • Structured data extraction

Usage

HuggingFace

pip install transformers torch
from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-pro-32b")
model = AutoModelForCausalLM.from_pretrained("zenlm/zen-pro-32b", device_map="auto")

inputs = tokenizer("Draft a technical proposal for:", return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

API

from hanzoai import Hanzo

client = Hanzo(api_key="hk-your-api-key")

response = client.chat.completions.create(
    model="zen-pro",
    messages=[{"role": "user", "content": "Analyze the trade-offs between microservices and monolithic architecture."}],
)
print(response.choices[0].message.content)

See Also

  • zen -- 8-32B standard foundation model
  • zen4-pro -- 80B MoE professional tier
  • zen-max -- 1.04T MoE frontier scale

On this page