WizardLM-2 8x22B
Microsoft's 141B mixture-of-experts chat model built on Mixtral 8x22B.
About
WizardLM-2 8x22B is a large mixture-of-experts language model released by Microsoft in April 2024, built on the Mixtral 8x22B architecture with roughly 141 billion total parameters. It was trained with Microsoft's WizardLM instruction-tuning pipeline, which emphasizes complex, multi-step instructions and agentic-style task completion.
The model offers a 65.5K token context window and ships with open weights on Hugging Face. It landed as one of the strongest open MoE chat models of early 2024, frequently benchmarked against contemporaries like DeepSeek and the GLM series. Exact license terms should be checked on the model card before production use.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 132; Trending score 78/100 from researcher assessment.