Using Zephyr ORPO 141B on DeepInfra

Quick Start

1. 1
   Create an account at DeepInfra and generate an API key.
2. 2
   Use the DeepInfra SDK or REST API to call zephyr-orpo-141b — see the documentation for request format.
3. 3
   You'll be billed $0.65/1M input, $0.65/1M output tokens. See full pricing.

Code Examples

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["DEEPINFRA_API_KEY"],
    base_url="https://api.deepinfra.com/v1/openai"
)
response = client.chat.completions.create(
    model="zephyr-orpo-141b",
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

About DeepInfra

DeepInfra offers serverless AI inference with a simple API, supporting hundreds of models across text generation, embeddings, and more. Pay-per-token pricing with no upfront commitments.

DeepInfra is a cloud inference platform offering cost-effective access to open-source AI models. It provides serverless inference for leading models from Meta, Mistral, Alibaba, and others with competitive token-based pricing.

Pricing on DeepInfra

Capabilities

About Zephyr ORPO 141B

The Zephyr ORPO 141B is a cutting-edge large language model by Hugging Face, developed in partnership with Argilla and KAIST. It employs a Mixture of Experts (MoE) architecture, consisting of 141 billion parameters, with 39 billion active during operation. The model is derived from the Mixtral-8x22B framework and fine-tuned using the innovative Odds Ratio Preference Optimization (ORPO) method, which improves computational efficiency by removing the need for a separate supervised fine-tuning phase. Zephyr ORPO demonstrates impressive proficiency in tasks such as open-ended conversations, question answering, and coding assistance, evidenced by high scores on benchmarks like MT Bench (8.17) and IFEval (65.06) 139. The model, trained on a dataset of 7,000 instances, is optimized for multi-turn dialogues, making it ideal for interactive AI applications. However, attention should be paid to its lack of alignment with human safety preferences, as this could lead to problematic outputs 4512.