FinFIRST
Agents research financial questions and provide supporting sources. The benchmark grades individual requirements and reports how often an answer meets every requirement.
- Made by
- Wenqing Wang and colleagues
- Released
- September 2026
- Kind
- Finance
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Subject
- Economics and business
- Language
- English
- Interaction
- Agentic
- Environment
- Browser
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents, Law and finance
- Strongest model tested
- Claude Opus 5 #11 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
GPT-5.6 Sol leads with 71.5% among models released by September 2026
Best score per model from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best.
The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- Claude Opus 5.5
- and 14 more
- 1 GPT-5.6 Sol 71.5%
- 2 Claude Opus 5 69.1%
- 3
Qwen3.8 Flash 61.8%
- 4 GLM-5.3 61%
- 5 Kimi K3 59.4%
- 6 GLM-5.3-Flash 56.9%
- 7
Qwen 3.8 27B 55.3%
- 8
Qwen 3.8 Max 54.5%
- 9
Ling-3.0-flash-Fin 52.9%
- 10
DeepSeek V4 Pro 0813 51.2%
- 11
DeepSeek V4 Flash 0731 48.8%
- 12
Gemini 3.7 Flash 44.7%
- 13 GLM-5.2 43.1%
- 14
Tencent Hy3 40.6%
- 15
MiniMax-M3 37.4%
Source: The FinFIRST paper
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.6 Sol | 71.5% | ReAct | Jul 2026 | Official leaderboard |
| 2 | Claude Opus 5 | 69.1% | ReAct | Jul 2026 | Official leaderboard |
| 3 | | 61.8% | ReAct | Aug 2026 | Official leaderboard |
| 4 | GLM-5.3 | 61% | ReAct | Aug 2026 | Official leaderboard |
| 5 | Kimi K3 | 59.4% | ReAct | Jul 2026 | Official leaderboard |
| 6 | GLM-5.3-Flash | 56.9% | ReAct | Aug 2026 | Official leaderboard |
| 7 | | 55.3% | ReAct | Aug 2026 | Official leaderboard |
| 8 | | 54.5% | ReAct | Aug 2026 | Official leaderboard |
| 9 | | 52.9% | ReAct | Sep 2026 | Official leaderboard |
| 10 | | 51.2% | ReAct | Aug 2026 | Official leaderboard |
| 11 | | 48.8% | ReAct | Jul 2026 | Official leaderboard |
| 12 | | 44.7% | ReAct | Aug 2026 | Official leaderboard |
| 13 | GLM-5.2 | 43.1% | ReAct | Jun 2026 | Official leaderboard |
| 14 | | 40.6% | ReAct | Jul 2026 | Official leaderboard |
| 15 | | 37.4% | ReAct | Jun 2026 | Official leaderboard |
Rankings by part
We read 3 parts from the benchmark’s paper: Raw-information, Source verification and Computation & answer. Each ranks models on one part of the test. Claude Opus 5 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.
Raw-information
15 models we track have a score here.
- 1 Claude Opus 5 ↑1 89%
- 2 GPT-5.6 Sol ↓1 86.8%
- 3 Kimi K3 ↑2 84.8%
- 4
Qwen3.8 Flash ↓1 84.3%
- 5 GLM-5.3 ↓1 83.1%
- 6
Qwen 3.8 27B ↑1 81.5%
- 7
Gemini 3.7 Flash ↑5 81.2%
- 8
Qwen 3.8 Max 80.3%
- 9
DeepSeek V4 Pro 0813 ↑1 80.1%
- 10 GLM-5.3-Flash ↓4 79.5%
Source verification
15 models we track have a score here.
- 1 Claude Opus 5 ↑1 89.6%
- 2 GPT-5.6 Sol ↓1 88.7%
- 3
Qwen3.8 Flash 83.4%
- 4
Ling-3.0-flash-Fin ↑5 82.5%
- 5 GLM-5.3-Flash ↑1 80.8%
- 6 GLM-5.3 ↓2 78.6%
- 7
Qwen 3.8 Max ↑1 77.8%
- 8
Qwen 3.8 27B ↓1 76.8%
- 9
DeepSeek V4 Pro 0813 ↑1 73.6%
- 10
DeepSeek V4 Flash 0731 ↑1 71.7%
Computation & answer
15 models we track have a score here.
- 1 Claude Opus 5 ↑1 82.7%
- 2 GPT-5.6 Sol ↓1 81.6%
- 3 Kimi K3 ↑2 78%
- 4 GLM-5.3 75.8%
- 5
Gemini 3.7 Flash ↑7 72.8%
- 6
Qwen3.8 Flash ↓3 71.9%
- 7
Qwen 3.8 Max ↑1 69.7%
- 8
Qwen 3.8 27B ↓1 66.9%
- 9 GLM-5.3-Flash ↓3 66.4%
- 10
DeepSeek V4 Pro 0813 64.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The FinFIRST paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official FinFIRST leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Law, finance and medicine
- MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy.
Qwen 3.8 Max 80.4%
- HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
- HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
- FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy.
Ling-3.1-flash 81.7%
- KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
- Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%
Data audit
All 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (15 entries).
- Pass Models scored 15 models we track.
- Pass Newest score The newest is from September 2026, for Ling-3.0-flash-Fin.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 5 (#11).
- Pass Headroom The best is 71.5%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 15 rows for models we track at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.