M

WizardLM-2 8x22B

Microsoft
128K
Context
$0.4800
Input /1M
$0.4800
Output /1M
16K
Max Output
Run API Call Open in Chat

WizardLM-2 8x22B

Overview

Property Value
Model Name WizardLM-2 8x22B
Model ID microsoft/wizardlm-2-8x22b
Creator Microsoft AI
Base Model Mixtral 8x22B (Mistral)
Architecture Mixture of Experts (MoE)
Release Date April 16, 2024

Description

Microsoft's most advanced Wizard model, described as demonstrating "highly competitive performance compared to leading proprietary models" and consistently outperforming existing open-source alternatives. It's an instruction-tuned version of Mixtral 8x22B.

Provider

  • Primary Provider: DeepInfra

Technical Specifications

Specification Value
Context Window 128,000 tokens
Context Length 65,536 tokens
Max Completion Tokens 16,384 tokens
Input Type Text only
Output Type Text only

Pricing

Type Price
Input $0.48 per 1M tokens
Output $0.48 per 1M tokens

Capabilities

  • Text input/output processing
  • Instruction following
  • Tool choice support (none, auto, required)
  • Abortable requests
  • High performance on reasoning tasks

Supported Parameters

Parameter Description
max_tokens Maximum number of tokens to generate
temperature Controls randomness (0-2)
top_p Nucleus sampling threshold
stop Stop sequences
frequency_penalty Reduces repetition of frequent tokens
presence_penalty Reduces repetition of any tokens
repetition_penalty Alternative repetition control
top_k Top-k sampling
seed Random seed for reproducibility
min_p Minimum probability threshold
response_format Output format specification

Performance

  • Demonstrates highly competitive performance compared to leading proprietary models
  • Consistently outperforms existing open-source alternatives
  • Actively used with significant daily token consumption (millions of tokens daily across thousands of requests)
  • Base Model: Mistral Mixtral 8x22B
  • Model Family: WizardLM series by Microsoft

Usage Example

curl -X POST https://api.langmart.ai/v1/chat/completions \
  -H "Authorization: Bearer $LANGMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "microsoft/wizardlm-2-8x22b",
    "messages": [
      {"role": "user", "content": "Your prompt here"}
    ],
    "max_tokens": 1024,
    "temperature": 0.7
  }'

Resources

  • Model weights available on Hugging Face
  • LangMart endpoint: https://langmart.ai/model-docs

Data sourced from LangMart on December 23, 2025