The Fin AI is a global academy advancing financial general intelligence through education, open research and incubation. This is where our code lives — evaluation frameworks, benchmarks, models and agent systems for financial AI. Datasets, model weights and live leaderboards are on Hugging Face.
Open-source, trustworthy Financial General Intelligence — pursued together by researchers, engineers and financial experts around the world. AI Readiness in Finance.
| 🎓 Education | Bringing and educating the next generation of financial AI researchers and talents through courses, study groups and hands-on learning. |
| 🔬 Research | Driving rigorous and transparent research with open benchmarks, models and agent systems that advance the science of financial AI. |
| 🚀 Incubation | Transforming ideas into impactful financial AI solutions by supporting early-stage teams with guidance, resources and real-world use cases. |
| Project | What it is | Paper | Data |
|---|---|---|---|
| PIXIU | FinMA models, instruction data and the FLARE evaluation suite | 2306.05443 | 🤗 |
| FinBen | Holistic financial benchmark for LLMs | 2402.12659 | 🤗 |
| MultiFinBen | Multilingual & multimodal financial evaluation | 2506.14028 | 🤗 |
| Ebisu | Japanese financial language evaluation | 2602.01479 | 🤗 |
| CALM | Credit scoring and risk assessment | 2310.00566 | 🤗 |
| FinTagging | XBRL tagging of financial reports | 2505.20650 | 🤗 |
| FinAuditing | Taxonomy-structured, multi-document auditing (SIGIR 2026) | 2510.08886 | 🤗 |
| FinCriticalED | Fact-level OCR of financial documents | 2511.14998 | 🤗 |
| Conv-FinRe | Conversational financial recommendation | 2602.16990 | 🤗 |
| Project | What it is | Paper |
|---|---|---|
| Fino1 | Reasoning-enhanced financial LLMs (8B / 14B) | 2502.08127 |
| FinSeer | Retrieval-augmented financial time-series forecasting | 2502.05878 |
| HARMONIC | LLMs for tabular data synthesis and privacy protection | 2408.02927 |
| CommitteeAudit | The "standing committee" in Mixture-of-Experts models (ACL 2026) | 2601.03425 |
| Project | What it is | Paper |
|---|---|---|
| Agent Market Arena | Live multi-market trading benchmark for LLM agents · live arena | 2510.11695 |
| FinCon | LLM multi-agent system for financial decision making (NeurIPS 2024) | 2407.06567 |
| trials | Pair trading with hierarchical reinforcement learning | 2301.10724 |
- Open FinLLM Leaderboard · Reasoning Leaderboard · Greek Financial LLM Leaderboard
- FinReason Cup 2026 — IEEE Big Data Cup: agentic financial reasoning, hedging & audit
- FinNLP 2026 — 11th Workshop on Financial Technology and NLP, co-located with EMNLP 2026
Questions or ideas? Open an issue on the relevant repository, or join us on Discord and Google Groups. More at thefin.ai · YouTube.