BFSI-Bench is a benchmark for testing how well language models answer questions about India’s banking, financial services, and insurance (BFSI) rules.

In this domain, the correct answer often depends on circulars and regulations that change frequently, and the official sources (sites like RBI, SEBI, and IRDAI) can be hard to find, parse, and keep current. BFSI-Bench measures five capability areas:

Jurisdiction-Aware Compliance: Disambiguate to the Indian context, or ask a clarifying question, instead of defaulting to U.S. or EU rules.
Numerical Reasoning: Get finance math right, including EMI, interest, and TDS.
Temporal Logic: Prefer currently applicable circulars and limits when the prompt does not specify a year.
Red-Team Evasion: Refuse illegal financial workarounds and point users to legitimate channels.
Groundedness: Stay faithful to a circular, scheme, or policy that the query names.

Questions and gold answers are expert-written and checked against official Indian sources. Browse 5 example items per category on the project page. Request access here for the full test set.


Name	`ground-truth/bfsi-bench`
Items	190
Languages	English
Split	`test`
License	CC BY 4.0
Site	ground-truth.in
Project page	ground-truth/bfsi-bench

Schema

Field	Description
`id`	Stable public id (`bfsi-jur-001`, `bfsi-num-001`, `bfsi-grd-001`, …)
`category`	Category name
`category_code`	Short code (`BFSI-JUR`, `BFSI-NUM`, `BFSI-TMP`, `BFSI-ADV`, or `BFSI-GRD`)
`query`	The question / user prompt
`gold_answer`	Expert-written prose answer
`gold_nuggets`	JSON list of atomic, independently checkable facts
`sources`	Source list as JSON
`evidence`	JSON list of supporting quotes (see below). Full source documents are not included yet.

Each sources item looks like:

Field	Description
`url`	Official source URL, or `null` for note-only cites
`note`	Optional human note / title
`published_date`	ISO `YYYY-MM-DD` issue/publication date of that source when known; `null` for note-only cites

Each gold_nuggets item looks like:

Field	Description
`id`	Stable within the item (`n1`, `n2`, …)
`text`	One checkable proposition
`vital`	`true` if this fact is required for a correct answer
`evidence_quote`	Verbatim source span when the nugget is a source fact

Each evidence item looks like:

Field	Description
`url`	Official source URL, or `null` for note-only evidence
`title`	Short source title
`quote`	supporting span from the source content
`context`	Broader retrieved passage around the quote
`kind`	`source` when a URL is present, otherwise `note`

Benchmark results

Models are evaluated without tools and without an India/BFSI system prompt, with only this default instruction:

Answer the user's question directly and keep your reply concise.

Each model answer is compared to the item's gold nuggets. The grader applies category-specific criteria (for example, jurisdiction handling on JUR items or safe refusal on ADV items) and may use supporting quotes from the dataset's official sources. Each item receives one of three verdicts: correct, incorrect, or not_attempted.

Scores below are parametric (no tools), nugget-v1, grader openai/gpt-5.6-luna, 190 test items.

Model	Overall	BFSI-JUR	BFSI-NUM	BFSI-TMP	BFSI-ADV	BFSI-GRD
`openai/gpt-5.6-sol`	86.32	77.46	100.0	66.67	93.1	94.44
`anthropic/claude-fable-5`	84.21	70.42	100.0	61.11	96.55	97.22
`meta/muse-spark-1.2`	77.37	56.34	94.44	61.11	93.1	97.22
`google/gemini-3.7-flash`	66.32	43.66	86.11	50.0	72.41	94.44

How to evaluate

1. Generate predictions

Create a JSONL file (for example answers.jsonl) with one JSON object per line. Each id must match the dataset (bfsi-jur-001, …).

{"id": "bfsi-jur-001", "answer": "...your model's answer..."}

2. Grade answers

hf download ground-truth/bfsi-bench --repo-type dataset --local-dir bfsi-bench

export OPENROUTER_API_KEY=...   # default grader: openai/gpt-5.6-luna

uv run bfsi-bench/eval/evaluate.py --predictions answers.jsonl --out-dir results/my-model

Smoke-test with --limit 5 or --per-category 1. Outputs: graded.jsonl, summary.csv, report.md.

To generate answers from an OpenAI-compatible endpoint and then grade:

uv run bfsi-bench/eval/evaluate.py \
  --model openai/gpt-4o \
  --api-base-url https://openrouter.ai/api/v1 \
  --limit 5

The grading rubric is eval/grading.py. A short run guide is in eval/README.md.

License

This dataset is released under CC BY 4.0. It is meant for evaluating language models on Indian BFSI questions.

Citation

@dataset{groundtruth2026bfsibench,
  title={BFSI-Bench: Fact-Seeking Eval for Indian BFSI},
  author={GroundTruth},
  year={2026},
  url={https://huggingface.co/datasets/ground-truth/bfsi-bench},
  license={CC-BY-4.0}
}

Contact

We plan to expand this into a continuous evaluation benchmark that stays up to date as Indian BFSI regulations evolve, while adding new evaluation categories such as multilingual capabilities, implicit groundedness, and false-presupposition testing.

Building evaluations for Indian BFSI, or interested in early access to upcoming GroundTruth datasets? Reach out at miroojin@ground-truth.in or visit ground-truth.in.

Downloads last month: 215

Total file size:

916 kB

Datasets:

ground-truth
/

bfsi-bench

You need to agree to share your contact information to access this dataset

BFSI-Bench

Categories