Mistral: Mixtral 8x7B Instruct
Mistral • text • function-calling • json-mode
Provider ID
mistralai/mixtral-8x7b-instructMixtral 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
Quick Summary
Best For:
High-volume, low-latency tasks where cost efficiency is paramount
Pricing:
$0.00/1M input tokens, $0.00/1M output tokens
Context Window:
32,768 tokens
Key Differentiator:
Cost-optimized for high-volume usage
Specifications
Context Window
32,768 tokens
Max Output Tokens
16,384 tokens
Streaming
Yes
JSON Mode
Yes
Vision
No
Tier
Affordable
Capabilities
text
function-calling
json-mode
Tags
affordable