Mistral: Mixtral 8x7B Instruct

Text input Text output
Author's Description

Mixtral 8x7B Instruct is a pretrained generative Sparse Mixture of Experts, by Mistral AI, for chat and instruction use. Incorporates 8 experts (feed-forward networks) for a total of 47 billion parameters. Instruct model fine-tuned by Mistral. #moe

Key Specifications
Cost
$$
Context
32K
Parameters
56B
Released
Dec 09, 2023
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Stop Tools Frequency Penalty Presence Penalty Top P Response Format Tool Choice Temperature Seed Min P Max Tokens
Features

This model supports the following features:

Response Format Tools
Performance Summary

Mistral's Mixtral 8x7B Instruct model consistently performs among the fastest models available, demonstrating exceptional speed across various benchmarks. It also offers highly competitive pricing, making it an economically attractive option. The model exhibits strong reliability with a 92% success rate, indicating few technical issues and consistent delivery of usable responses. In terms of performance across categories, Mixtral 8x7B Instruct shows notable strengths in Ethics and General Knowledge, achieving 99.5% and 96.8% accuracy respectively. Its performance in Hallucinations is fair at 84.0% accuracy, suggesting it generally acknowledges uncertainty appropriately. A significant weakness is observed in Instruction Following, where it scored 0.0% accuracy, indicating a critical area for improvement. Performance in Reasoning (34.0% accuracy) and Coding (38.0% accuracy) is also relatively low. While its Email Classification accuracy is 82.0%, it stands out as one of the fastest models in this category. Overall, the model excels in foundational knowledge and ethical reasoning, but struggles with complex instruction following, reasoning, and coding tasks.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.54
Completion $0.54

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
DeepInfra
DeepInfra | mistralai/mixtral-8x7b-instruct 32K $0.54 / 1M tokens $0.54 / 1M tokens
Together
Together | mistralai/mixtral-8x7b-instruct 32K $0.6 / 1M tokens $0.6 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by mistralai