MorbidCorp/Morbid-22B-Insurance-v020

Morbid v0.2.0 - Enterprise Insurance AI

A 22B parameter LLM fine-tuned for health and life insurance applications, built on Mistral Small Instruct.

Model Details

• Base Model: mistralai/Mistral-Small-Instruct-2409
• Parameters: 22B
• Training: Supervised Fine-Tuning (SFT) with LoRA on insurance/actuarial dataset
• License: Apache 2.0
• Developed by: MorbidCorp

Capabilities

Insurance & Actuarial

• Life insurance products (term, whole, universal, variable)
• Health insurance (medical, dental, disability, LTC)
• Premium calculations and rate setting
• Underwriting and risk classification
• Claims analysis and management
• Regulatory compliance (NAIC, state/federal)

Actuarial Mathematics

• Mortality tables and life expectancy calculations
• Present value and annuity calculations
• Reserve valuation
• Risk assessment and modeling

Medical Classification

• ICD-10 code lookup and explanation
• Cause-of-death classification
• Medical terminology

Usage

Basic Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "MorbidCorp/Morbid-22B-Insurance-v020"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

# Mistral Instruct format
system = """You are Morbi, an expert AI assistant specializing in health and life insurance, actuarial science, and risk analysis."""

user_msg = "What is the life expectancy for a 50-year-old male in the US?"

prompt = f"<s>[INST] {system}\n\n{user_msg} [/INST]"

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7, top_p=0.9)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

With Transformers Pipeline

from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="MorbidCorp/Morbid-22B-Insurance-v020",
    torch_dtype="bfloat16",
    device_map="auto"
)

messages = [
    {"role": "user", "content": "Explain the difference between term and whole life insurance."}
]

output = pipe(messages, max_new_tokens=512)
print(output[0]["generated_text"][-1]["content"])

Training Data

The model was fine-tuned on a curated dataset including:

• Insurance product documentation and explanations
• Actuarial exam questions and solutions (SOA P, FM, IFM)
• Life expectancy and mortality data
• ICD-10 WHO 2019 medical classification codes
• Underwriting guidelines and risk assessment scenarios
• Regulatory compliance documentation

Limitations

• This model is for informational purposes only
• Not a substitute for licensed professional advice
• Should not be used for final underwriting decisions
• May not reflect the most current regulatory requirements
• Life expectancy estimates are population averages, not individual predictions

Hardware Requirements

• Minimum: 40GB VRAM (A100 40GB, A6000)
• Recommended: 80GB VRAM (A100 80GB, H100)
• Quantized (AWQ/GPTQ 4-bit): 24GB VRAM (RTX 4090, A10G)

Citation

@misc{morbid2026,
  title={Morbid v0.2.0: Enterprise Insurance AI},
  author={MorbidCorp},
  year={2026},
  publisher={HuggingFace},
  url={https://huggingface.co/MorbidCorp/Morbid-22B-Insurance-v020}
}

Contact

For enterprise support and customization: MORBID.AI