Ministral-3-3B-Instruct-2512

Ministral-3-3B-Instruct-2512 (Mistral AI)

Compact instruction-tuned language model optimized for efficiency and reliability.

• Small but capable. A ~3B-parameter model from the Mistral family, tuned for instruction following, chat, and structured text tasks without the overhead of large LLMs.

• Clear instruction handling. Produces concise, well-structured answers and follows prompts more reliably than older small models.

• Balanced performance. Strong at summarization, classification, lightweight reasoning, and assistant workflows while keeping latency and memory usage low.

• Easy to deploy. Fits on modest GPUs or optimized CPU setups, making it suitable for high-throughput or cost-sensitive use cases.

• Open and practical. Released under a permissive license and integrates cleanly into standard LLM pipelines.

Why pick it for Norman AI?

Ministral-3-3B-Instruct-2512 is a solid middle option when you want dependable instruction following without scaling to large models. It works well for assistants, automation, and text processing where speed and cost matter.

messages = [
    {"role": "system", "content": "You are a helpful AI assistant."},
    {"role": "user", "content": "Can you provide ways to eat combinations of bananas and dragonfruits?"},
    {"role": "assistant",
     "content": "Sure! Here are some ways to eat bananas and dragonfruits together"},
    {"role": "user", "content": "What about solving an 2x + 3 = 7 equation?"},
]

response = await norman.invoke(
    {
        "model_name": "ministral-3-3b-instruct-2512",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": messages
            }
        ]
    }
)