Mixtral 8x22B is an LLM that excels in multilingual understanding, math reasoning, and code generation. It offers a high performance-to-cost ratio due to its sparse mixture-of-experts architecture, which allows for efficient processing with fewer active parameters.

Some other noteworthy features of Mixtral 8x22B include native function calling support and a large context window for improved information recall from extensive documents.

Metric	Value
Parameter Count	141 billion
Mixture of Experts	Yes
Active Parameter Count	39 billion
Context Length	64,000 tokens
Multilingual	Yes
Quantized*	No

*Quantization is specific to the inference provider and the model may be offered with different quantization levels by other providers.