FinnAI SLM v2

On-device model for Indian bank SMS → JSON and personal finance Q&A.

Useful when you want structured spends from bank SMS on the phone, without uploading messages to a cloud LLM.


Model	`finndot/finnai-slm-v2`
Dataset	`finndot/finnai-slm-data`
Base	`Qwen/Qwen3-1.7B`
License	Apache 2.0
App	FinnDot

FinnAI model family

Version	Weights	On-device LiteRT	Data
v4 (latest)	`finnai-slm-v4`	`finnai-slm-v4-litertlm`	`finnai-slm-data-v4`
v3	`finnai-slm-v3`	`finnai-slm-v3-litertlm`	`finnai-slm-data-v3`
v2 (this)	`finnai-slm-v2`	—	`finnai-slm-data`

Prefer v4 for new projects.

What it helps with

Capability	How it helps
SMS → JSON	Extracts amount, merchant, type, account, balance, category from Indian bank SMS
Noise refusal	Returns `{}` for OTP / promo / KYC instead of inventing spends
Finance coach	Short answers about spending, SIP, EMI, budgeting
Privacy	Designed for on-device expense tracking (FinnDot)

Quick start

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "finndot/finnai-slm-v2"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype="bfloat16", device_map="auto"
)

messages = [
    {"role": "system", "content": "Extract transaction details from this SMS as JSON: {amount, merchant, type, account, balance, category}. type is one of: INCOME, EXPENSE, CREDIT, TRANSFER, INVESTMENT. If this is not a transaction (OTP, promo, KYC), return {}."},
    {"role": "user", "content": "HDFC Bank: Rs.499.00 debited from A/c XX1234 to SWIGGY via UPI. Avl Bal Rs.15000.50"}
]
text = tokenizer.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True, enable_thinking=False
)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
output = model.generate(**inputs, max_new_tokens=256, do_sample=False)
print(tokenizer.decode(output[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Quality snapshot

Metric	FinnAI v2	Untuned Qwen3-1.7B
Strict R-EM %	97.97	~28
Amount EM %	99.53	high but over-parses noise
JSON valid %	100.0	—

Limitations

Indian bank SMS focus; may not fit US/EU formats well
Merchant field is hardest; regex parsers can still win on supported banks
Smaller than general assistants — built for SMS + short finance chat

Citation

@misc{finnai-slm-v2,
  title={FinnAI SLM: On-device Indian bank SMS parsing},
  author={FinnDot Team},
  year={2026},
  url={https://huggingface.co/finndot/finnai-slm-v2}
}

License

Apache 2.0 — same as Qwen3-1.7B.