What are the key features that distinguish Ministral 8B?
Ministral 8B features a special interleaved sliding-window attention pattern for faster and memory-efficient inference, along with support for up to 128k context length (currently 32k on vLLM).

How does Ministral 8B perform compared to other models?
It sets new state-of-the-art benchmarks in the sub-10B category for knowledge, commonsense, reasoning, and function-calling capabilities.

What is the pricing structure for Ministral 8B?
The API pricing on la Plateforme is $0.1 per million tokens (both input and output).

Is Ministral 8B available for research use?
Yes, the model weights for Ministral 8B Instruct are available for research use under the Mistral Research License.