Get Started with Mixtral 8x22B v0.1 on Microsoft Foundry
Microsoft Foundry offers access to Mixtral 8x22B v0.1 with a 64K context window. Microsoft Foundry is a unified enterprise AI platform that significantly expands beyond Azure OpenAI. It functions as a multi-provider hosting and deployment platform for LLMs, supporting models from OpenAI, Anthropic, DeepSeek, xAI, Meta, Mistral, NVIDIA, and others. Foundry integrates agent services, evaluation, observability, and governance into a single Azure control plane. Key capabilities include a multi-provider model catalog, Model Router for intelligent prompt routing, Foundry Agent Service for building and deploying AI agents with built-in tracing and monitoring, and enterprise-grade governance with RBAC, compliance, and regional deployments. For broader model catalog including Claude, DeepSeek, Grok, Llama, Mistral, and NVIDIA Nemotron, Foundry is the recommended platform over Azure OpenAI.
Pricing
| Type | Price (per 1M) |
|---|---|
| Input tokens | $2.00 |
| Output tokens | $6.00 |
Capabilities
About Mixtral 8x22B v0.1
The Mixtral 8x22B v0.1 is a pretrained generative Sparse Mixture of Experts (MoE) model created by Mistral AI [1][2][4]. It utilizes a specialized architecture where different sub-models, termed "experts," manage distinct input segments, enhancing both efficiency and performance relative to traditional large language models [2][10][12]. This model features an impressive 176 billion parameters and supports a context length of 65,000 tokens [10][13]. It excels in text generation, completion, and question answering, outperforming models like LLaMA 2 70B on various benchmarks [4][5][7]. Nonetheless, as a base model, it lacks inherent moderation capabilities, potentially generating inappropriate or harmful content without filtration [2][4][10]. The model requires significant VRAM—approximately 260GB in FP16 mode and 73GB in INT4 mode—for optimal operation [10][13] and may struggle with complex contextual understanding and current knowledge. Enhanced instruct-tuned versions, such as the Mixtral-8x22B-Instruct-v0.1, address some limitations by improving instruction adherence [3][5][6].