FinGREAT
Models identify accounting relationships in financial passages. The authors compare 44 models against human annotations across 483 passages.
- Made by
- Carnegie Mellon University
- Released
- October 2026
- Kind
- Finance
- Level
- University
- Scored
- Checked automatically
- Answers
- Short answer
- Measures
- Capability
- Subject
- Economics and business
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- In views
- Law and finance
- Strongest model tested
- GPT-6 Astra #4 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
2 models share the top score of 92.8% among models released by September 2026 2 models share the top score of 92.8% among models released by September 2026
Best score per entry 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. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at 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
- GPT-6.1 Terra
- and 11 more
- 92.8 Claude Fable 5.1
- 92.8 GPT-6 Astra
- 92.5 GPT-6.1 Sol
- 92.1 Claude Opus 5.5
- 91.3
Gemini 3.8 Flash
- 91.3 Claude Sonnet 5.5
- 89.6
Gemini 3.1 Pro
- 84.5 Claude Haiku 4.5
- 84.1
Qwen3.5-27B
- 84.1
Qwen3.6 27B
- 83.2
Gemma 4 12B
- 83.0
Gemma 4 31B IT
- 82.4
Gemini 3.5 Flash-Lite
- 80.1 Ministral-3-14B-Instruct
- 79.9
Gemma 4 26B A4B
- 79.3 GPT-6 Luna
- 78.7
Qwen3.5-9B
- 78.5
Qwen3.5-35B-A3B
- 77.0
Qwen 3.6 35B-A3B
- 74.9
Gemma 3 27B
- 74.5
Gemma 4 E4B
- 71.2
Gemma 3 12B
- 70.2
Phi-4
- 69.2
Qwen3.5-4B
- 66.9
Ministral 3 3B Instruct
- 65.2
Gemma 4 E2B
- 64.8
Ministral 3 8B Instruct
- 58.8 Phi-3-Small-128K-Instruct
- 54.7
Phi-3.5-mini
- 54.2
Gemma 3 4B
- 52.8 Llama-3.1-8B-Instruct
- 49.5
Phi-3-mini-128k-instruct
- 48.0 Llama-3.2-3B-Instruct
- 46.6 Meta-Llama-3-8B-Instruct
- 45.1
Phi-4 Mini
- 42.0 DeepSeek-V2-Lite-Chat
- 41.8
Qwen3.5-2B
- 36.4
Qwen3.5-0.8B
- 23.0 Llama-3.2-1B-Instruct
- 22.4
Gemma 3 1B
- 20.9 DeepSeek-LLM-7B-Chat
- 19.5
Phi-3-medium-128k-instruct
- 15.9 DeepSeek-MoE-16B-Chat
- 8.90
Gemma 3 270M
- 1 Claude Fable 5.1 92.8%
- 2 GPT-6 Astra 92.8%
- 3 GPT-6.1 Sol 92.5%
- 4 Claude Opus 5.5 92.1%
- 5
Gemini 3.8 Flash 91.3%
- 6 Claude Sonnet 5.5 91.3%
- 7
Gemini 3.1 Pro 89.6%
- 8 Claude Haiku 4.5 84.5%
- 9
Qwen3.5-27B 84.1%
- 10
Qwen3.6 27B 84.1%
- 11
Gemma 4 12B 83.2%
- 12
Gemma 4 31B IT 83%
- 13
Gemini 3.5 Flash-Lite 82.4%
- 14 Ministral-3-14B-Instruct 80.1%
- 15
Gemma 4 26B A4B 79.9%
- 16 GPT-6 Luna 79.3%
- 17
Qwen3.5-9B 78.7%
- 18
Qwen3.5-35B-A3B 78.5%
- 19
Qwen 3.6 35B-A3B 77%
- 20
Gemma 3 27B 74.9%
- 21
Gemma 4 E4B 74.5%
- 22
Gemma 3 12B 71.2%
- 23
Phi-4 70.2%
- 24
Qwen3.5-4B 69.2%
- 25
Ministral 3 3B Instruct 66.9%
- 26
Gemma 4 E2B 65.2%
- 27
Ministral 3 8B Instruct 64.8%
- 28 Phi-3-Small-128K-Instruct 58.8%
- 29
Phi-3.5-mini 54.7%
- 30
Gemma 3 4B 54.2%
- 31 Llama-3.1-8B-Instruct 52.8%
- 32
Phi-3-mini-128k-instruct 49.5%
- 33 Llama-3.2-3B-Instruct 48%
- 34 Meta-Llama-3-8B-Instruct 46.6%
- 35
Phi-4 Mini 45.1%
- 36 DeepSeek-V2-Lite-Chat 42%
- 37
Qwen3.5-2B 41.8%
- 38
Qwen3.5-0.8B 36.4%
- 39 Llama-3.2-1B-Instruct 23%
- 40
Gemma 3 1B 22.4%
- 41 DeepSeek-LLM-7B-Chat 20.9%
- 42
Phi-3-medium-128k-instruct 19.5%
- 43 DeepSeek-MoE-16B-Chat 15.9%
- 44
Gemma 3 270M 8.90%
Source: The FinGREAT paper
- 92.8 Claude Fable 5.1
- 92.8 GPT-6 Astra
- 92.5 GPT-6.1 Sol
- 92.1 Claude Opus 5.5
- 91.3
Gemini 3.8 Flash
- 91.3 Claude Sonnet 5.5
- 89.6
Gemini 3.1 Pro
- 84.5 Claude Haiku 4.5
- 84.1
Qwen3.5-27B
- 84.1
Qwen3.6 27B
- 83.2
Gemma 4 12B
- 83.0
Gemma 4 31B IT
- 82.4
Gemini 3.5 Flash-Lite
- 79.9
Gemma 4 26B A4B
- 79.3 GPT-6 Luna
- 78.7
Qwen3.5-9B
- 78.5
Qwen3.5-35B-A3B
- 77.0
Qwen 3.6 35B-A3B
- 74.9
Gemma 3 27B
- 74.5
Gemma 4 E4B
- 71.2
Gemma 3 12B
- 70.2
Phi-4
- 69.2
Qwen3.5-4B
- 66.9
Ministral 3 3B Instruct
- 65.2
Gemma 4 E2B
- 64.8
Ministral 3 8B Instruct
- 54.7
Phi-3.5-mini
- 54.2
Gemma 3 4B
- 49.5
Phi-3-mini-128k-instruct
- 45.1
Phi-4 Mini
- 41.8
Qwen3.5-2B
- 36.4
Qwen3.5-0.8B
- 22.4
Gemma 3 1B
- 19.5
Phi-3-medium-128k-instruct
- 8.90
Gemma 3 270M
- 1 Claude Fable 5.1 92.8%
- 2 GPT-6 Astra 92.8%
- 3 GPT-6.1 Sol 92.5%
- 4 Claude Opus 5.5 92.1%
- 5
Gemini 3.8 Flash 91.3%
- 6 Claude Sonnet 5.5 91.3%
- 7
Gemini 3.1 Pro 89.6%
- 8 Claude Haiku 4.5 84.5%
- 9
Qwen3.5-27B 84.1%
- 10
Qwen3.6 27B 84.1%
- 11
Gemma 4 12B 83.2%
- 12
Gemma 4 31B IT 83%
- 13
Gemini 3.5 Flash-Lite 82.4%
- 14
Gemma 4 26B A4B 79.9%
- 15 GPT-6 Luna 79.3%
- 16
Qwen3.5-9B 78.7%
- 17
Qwen3.5-35B-A3B 78.5%
- 18
Qwen 3.6 35B-A3B 77%
- 19
Gemma 3 27B 74.9%
- 20
Gemma 4 E4B 74.5%
- 21
Gemma 3 12B 71.2%
- 22
Phi-4 70.2%
- 23
Qwen3.5-4B 69.2%
- 24
Ministral 3 3B Instruct 66.9%
- 25
Gemma 4 E2B 65.2%
- 26
Ministral 3 8B Instruct 64.8%
- 27
Phi-3.5-mini 54.7%
- 28
Gemma 3 4B 54.2%
- 29
Phi-3-mini-128k-instruct 49.5%
- 30
Phi-4 Mini 45.1%
- 31
Qwen3.5-2B 41.8%
- 32
Qwen3.5-0.8B 36.4%
- 33
Gemma 3 1B 22.4%
- 34
Phi-3-medium-128k-instruct 19.5%
- 35
Gemma 3 270M 8.90%
Source: The FinGREAT paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 | 92.8% | Sep 2026 | Official leaderboard | |
| 2 | GPT-6 Astra | 92.8% | Sep 2026 | Official leaderboard | |
| 3 | GPT-6.1 Sol | 92.5% | Sep 2026 | Official leaderboard | |
| 4 | Claude Opus 5.5 | 92.1% | Sep 2026 | Official leaderboard | |
| 5 | | 91.3% | Sep 2026 | Official leaderboard | |
| 6 | Claude Sonnet 5.5 | 91.3% | Sep 2026 | Official leaderboard | |
| 7 | | 89.6% | Feb 2026 | Official leaderboard | |
| 8 | Claude Haiku 4.5 | 84.5% | Oct 2025 | Official leaderboard | |
| 9 | | 84.1% | Feb 2026 | Official leaderboard | |
| 10 | | 84.1% | Apr 2026 | Official leaderboard | |
| 11 | | 83.2% | Jun 2026 | Official leaderboard | |
| 12 | | 83% | Apr 2026 | Official leaderboard | |
| 13 | | 82.4% | Jul 2026 | Official leaderboard | |
| 14 | Ministral-3-14B-Instruct | 80.1% | Official leaderboard | ||
| 15 | | 79.9% | Apr 2026 | Official leaderboard | |
| 16 | GPT-6 Luna | 79.3% | Sep 2026 | Official leaderboard | |
| 17 | | 78.7% | Feb 2026 | Official leaderboard | |
| 18 | | 78.5% | Feb 2026 | Official leaderboard | |
| 19 | | 77% | Apr 2026 | Official leaderboard | |
| 20 | | 74.9% | Mar 2025 | Official leaderboard | |
| 21 | | 74.5% | Mar 2026 | Official leaderboard | |
| 22 | | 71.2% | Mar 2025 | Official leaderboard | |
| 23 | | 70.2% | Dec 2024 | Official leaderboard | |
| 24 | | 69.2% | Feb 2026 | Official leaderboard | |
| 25 | | 66.9% | Dec 2025 | Official leaderboard | |
| 26 | | 65.2% | Mar 2026 | Official leaderboard | |
| 27 | | 64.8% | Dec 2025 | Official leaderboard | |
| 28 | Phi-3-Small-128K-Instruct | 58.8% | Official leaderboard | ||
| 29 | | 54.7% | Aug 2024 | Official leaderboard | |
| 30 | | 54.2% | Mar 2025 | Official leaderboard | |
| 31 | Llama-3.1-8B-Instruct | 52.8% | Official leaderboard | ||
| 32 | | 49.5% | Apr 2024 | Official leaderboard | |
| 33 | Llama-3.2-3B-Instruct | 48% | Official leaderboard | ||
| 34 | Meta-Llama-3-8B-Instruct | 46.6% | Official leaderboard | ||
| 35 | | 45.1% | Feb 2025 | Official leaderboard | |
| 36 | DeepSeek-V2-Lite-Chat | 42% | Official leaderboard | ||
| 37 | | 41.8% | Feb 2026 | Official leaderboard | |
| 38 | | 36.4% | Feb 2026 | Official leaderboard | |
| 39 | Llama-3.2-1B-Instruct | 23% | Official leaderboard | ||
| 40 | | 22.4% | Mar 2025 | Official leaderboard | |
| 41 | DeepSeek-LLM-7B-Chat | 20.9% | Official leaderboard | ||
| 42 | | 19.5% | May 2024 | Official leaderboard | |
| 43 | DeepSeek-MoE-16B-Chat | 15.9% | Official leaderboard | ||
| 44 | | 8.90% | Aug 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 | 92.8% | Sep 2026 | Official leaderboard | |
| 2 | GPT-6 Astra | 92.8% | Sep 2026 | Official leaderboard | |
| 3 | GPT-6.1 Sol | 92.5% | Sep 2026 | Official leaderboard | |
| 4 | Claude Opus 5.5 | 92.1% | Sep 2026 | Official leaderboard | |
| 5 | | 91.3% | Sep 2026 | Official leaderboard | |
| 6 | Claude Sonnet 5.5 | 91.3% | Sep 2026 | Official leaderboard | |
| 7 | | 89.6% | Feb 2026 | Official leaderboard | |
| 8 | Claude Haiku 4.5 | 84.5% | Oct 2025 | Official leaderboard | |
| 9 | | 84.1% | Feb 2026 | Official leaderboard | |
| 10 | | 84.1% | Apr 2026 | Official leaderboard | |
| 11 | | 83.2% | Jun 2026 | Official leaderboard | |
| 12 | | 83% | Apr 2026 | Official leaderboard | |
| 13 | | 82.4% | Jul 2026 | Official leaderboard | |
| 14 | | 79.9% | Apr 2026 | Official leaderboard | |
| 15 | GPT-6 Luna | 79.3% | Sep 2026 | Official leaderboard | |
| 16 | | 78.7% | Feb 2026 | Official leaderboard | |
| 17 | | 78.5% | Feb 2026 | Official leaderboard | |
| 18 | | 77% | Apr 2026 | Official leaderboard | |
| 19 | | 74.9% | Mar 2025 | Official leaderboard | |
| 20 | | 74.5% | Mar 2026 | Official leaderboard | |
| 21 | | 71.2% | Mar 2025 | Official leaderboard | |
| 22 | | 70.2% | Dec 2024 | Official leaderboard | |
| 23 | | 69.2% | Feb 2026 | Official leaderboard | |
| 24 | | 66.9% | Dec 2025 | Official leaderboard | |
| 25 | | 65.2% | Mar 2026 | Official leaderboard | |
| 26 | | 64.8% | Dec 2025 | Official leaderboard | |
| 27 | | 54.7% | Aug 2024 | Official leaderboard | |
| 28 | | 54.2% | Mar 2025 | Official leaderboard | |
| 29 | | 49.5% | Apr 2024 | Official leaderboard | |
| 30 | | 45.1% | Feb 2025 | Official leaderboard | |
| 31 | | 41.8% | Feb 2026 | Official leaderboard | |
| 32 | | 36.4% | Feb 2026 | Official leaderboard | |
| 33 | | 22.4% | Mar 2025 | Official leaderboard | |
| 34 | | 19.5% | May 2024 | Official leaderboard | |
| 35 | | 8.90% | Aug 2025 | Official leaderboard |
Rankings by part
We read 14 parts from the benchmark’s paper, each ranking models on one part of the test, such as Finance, Investment and Operations. Claude Fable 5.1 leads 5 of them, and 9 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Finance
35 models we track have a score here.
- 1
Qwen3.5-2B ↑30 89%
- 2
Qwen3.5-0.8B ↑30 86.6%
- 3 Claude Fable 5.1 ↓2 84.1%
- 4 GPT-6 Astra ↓2 84.1%
- 5 Claude Opus 5.5 ↓1 84.1%
- 6 GPT-6.1 Sol ↓3 84.1%
- 7 Claude Sonnet 5.5 ↓1 80.5%
- 8
Gemini 3.8 Flash ↓3 79.3%
- 9
Gemini 3.1 Pro ↓2 78%
- 10 Claude Haiku 4.5 ↓2 72%
Investment
35 models we track have a score here.
- 1 Claude Fable 5.1 92.9%
- 2 GPT-6.1 Sol ↑1 92.9%
- 3
Gemini 3.8 Flash ↑2 91.4%
- 4 GPT-6 Astra ↓2 91.4%
- 5 Claude Sonnet 5.5 ↑1 91.4%
- 6
Gemma 4 31B IT ↑6 88.6%
- 7 Claude Opus 5.5 ↓3 88.6%
- 8
Gemma 4 12B ↑3 87.1%
- 9
Gemini 3.1 Pro ↓2 84.3%
- 10
Ministral 3 3B Instruct ↑14 80%
Operations
35 models we track have a score here.
- 1 Claude Fable 5.1 97%
- 2
Gemini 3.8 Flash ↑3 93.9%
- 3 GPT-6 Astra ↓1 93.9%
- 4 Claude Opus 5.5 93.9%
- 5 GPT-6.1 Sol ↓2 93.9%
- 6
Gemma 4 31B IT ↑6 92.4%
- 7 Claude Sonnet 5.5 ↓1 90.9%
- 8 Claude Haiku 4.5 89.4%
- 9
Qwen3.6 27B ↑1 89.4%
- 10
Qwen3.5-9B ↑6 87.9%
Cash flow overall
35 models we track have a score here.
- 1 Claude Fable 5.1 89.7%
- 2 GPT-6.1 Sol ↑1 89.2%
- 3 GPT-6 Astra ↓1 88.7%
- 4 Claude Opus 5.5 88.2%
- 5 Claude Sonnet 5.5 ↑1 87.7%
- 6
Gemini 3.8 Flash ↓1 86.7%
- 7
Gemini 3.1 Pro 82.1%
- 8 Claude Haiku 4.5 78.5%
- 9
Gemma 4 31B IT ↑3 75.4%
- 10
Qwen3.6 27B 75.4%
Revenue
35 models we track have a score here.
- 1 Claude Fable 5.1 81.8%
- 2 Claude Opus 5.5 ↑2 81.8%
- 3
Gemini 3.1 Pro ↑4 72.7%
- 4 GPT-6 Astra ↓2 72.7%
- 5 GPT-6 Luna ↑10 72.7%
- 6 GPT-6.1 Sol ↓3 72.7%
- 7 Claude Haiku 4.5 ↑1 54.5%
- 8
Qwen3.5-27B ↑1 54.5%
- 9
Gemma 4 26B A4B ↑5 54.5%
- 10
Qwen3.6 27B 54.5%
Expenses
35 models we track have a score here.
- 1
Gemini 3.1 Pro ↑6 95.8%
- 2
Gemma 4 E2B ↑23 94.4%
- 3 Claude Opus 5.5 ↑1 94.4%
- 4 GPT-6.1 Sol ↓1 94.4%
- 5 Claude Fable 5.1 ↓4 93.1%
- 6 GPT-6 Astra ↓4 93.1%
- 7
Gemma 3 27B ↑12 91.7%
- 8 Claude Sonnet 5.5 ↓2 91.7%
- 9
Gemini 3.8 Flash ↓4 90.3%
- 10
Qwen3.5-35B-A3B ↑7 88.9%
Interest Expenses
35 models we track have a score here.
- 1
Gemma 4 E2B ↑24 100%
- 2
Phi-4 ↑20 97.3%
- 3
Qwen3.5-35B-A3B ↑14 97.3%
- 4
Gemma 4 12B ↑7 97.3%
- 5 Claude Sonnet 5.5 ↑1 97.3%
- 6
Gemma 3 12B ↑15 94.6%
- 7
Gemma 3 4B ↑21 94.6%
- 8
Ministral 3 3B Instruct ↑16 94.6%
- 9
Gemini 3.1 Pro ↓2 94.6%
- 10
Qwen3.5-27B ↓1 94.6%
Tax
35 models we track have a score here.
- 1
Gemma 3 27B ↑18 98.3%
- 2
Gemma 3 12B ↑19 98.3%
- 3
Gemma 4 12B ↑8 98.3%
- 4
Phi-4 ↑18 96.7%
- 5
Ministral 3 3B Instruct ↑19 96.7%
- 6
Gemini 3.1 Pro ↑1 96.7%
- 7
Qwen3.5-27B ↑2 96.7%
- 8
Qwen3.5-35B-A3B ↑9 96.7%
- 9
Qwen3.5-4B ↑14 96.7%
- 10
Qwen3.5-9B ↑6 96.7%
Net income overall
35 models we track have a score here.
- 1 Claude Fable 5.1 94.8%
- 2
Gemini 3.1 Pro ↑5 94.3%
- 3 Claude Opus 5.5 ↑1 94.3%
- 4 GPT-6.1 Sol ↓1 93.7%
- 5 GPT-6 Astra ↓3 93.1%
- 6
Qwen3.5-35B-A3B ↑11 92%
- 7
Gemma 4 E2B ↑18 92%
- 8 Claude Sonnet 5.5 ↓2 92%
- 9
Qwen3.6 27B ↑1 91.4%
- 10
Gemma 4 12B ↑1 91.4%
OCI
35 models we track have a score here.
- 1
Phi-4 ↑21 100%
- 2 Claude Haiku 4.5 ↑6 100%
- 3
Qwen3.5-27B ↑6 100%
- 4
Qwen3.5-35B-A3B ↑13 100%
- 5
Qwen3.5-4B ↑18 100%
- 6
Gemma 4 E4B ↑14 100%
- 7
Gemma 4 26B A4B ↑7 100%
- 8
Qwen3.6 27B ↑2 100%
- 9
Gemma 4 12B ↑2 100%
- 10
Gemini 3.5 Flash-Lite ↑3 100%
Dividends
35 models we track have a score here.
- 1 Claude Haiku 4.5 ↑7 97.7%
- 2
Qwen3.5-35B-A3B ↑15 97.7%
- 3
Qwen 3.6 35B-A3B ↑15 97.7%
- 4 GPT-6 Astra ↓2 97.7%
- 5
Phi-4 ↑17 95.5%
- 6
Qwen3.5-27B ↑3 95.5%
- 7
Gemma 4 E4B ↑13 95.5%
- 8
Gemma 4 26B A4B ↑6 95.5%
- 9
Gemma 4 31B IT ↑3 95.5%
- 10
Gemma 4 12B ↑1 95.5%
Non-cash Equity
35 models we track have a score here.
- 1
Gemini 3.8 Flash ↑4 100%
- 2 GPT-6 Astra 96.6%
- 3 Claude Opus 5.5 ↑1 96.6%
- 4 GPT-6.1 Sol ↓1 96.6%
- 5
Gemini 3.5 Flash-Lite ↑8 89.7%
- 6
Ministral 3 8B Instruct ↑20 86.2%
- 7
Gemini 3.1 Pro 86.2%
- 8 Claude Fable 5.1 ↓7 86.2%
- 9 Claude Sonnet 5.5 ↓3 79.3%
- 10
Gemma 4 E2B ↑15 75.9%
Cash Equity
35 models we track have a score here.
- 1
Qwen3.5-35B-A3B ↑16 99%
- 2 GPT-6 Astra 99%
- 3
Qwen3.5-27B ↑6 98.1%
- 4
Qwen 3.6 35B-A3B ↑14 98.1%
- 5
Qwen3.6 27B ↑5 97.1%
- 6 Claude Opus 5.5 ↓2 97.1%
- 7 GPT-6.1 Sol ↓4 97.1%
- 8
Gemma 4 26B A4B ↑6 96.1%
- 9
Gemma 4 31B IT ↑3 96.1%
- 10 Claude Fable 5.1 ↓9 96.1%
Equity overall
35 models we track have a score here.
- 1 GPT-6 Astra ↑1 97.7%
- 2
Gemini 3.8 Flash ↑3 95.9%
- 3 Claude Opus 5.5 ↑1 95.3%
- 4 GPT-6.1 Sol ↓1 95.3%
- 5 Claude Fable 5.1 ↓4 94.2%
- 6 Claude Sonnet 5.5 93.6%
- 7
Gemini 3.1 Pro 91.9%
- 8
Qwen3.5-27B ↑1 91.3%
- 9
Qwen3.6 27B ↑1 89.5%
- 10
Gemini 3.5 Flash-Lite ↑3 89%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The FinGREAT paper
Gemini 3.1 Pro on the official leaderboard passed 85% in February 2026
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official FinGREAT 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
9 of 10 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark
Catalog record 1 warning 1 pass
- Warning Task access We don’t know whether the tasks are public.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (44 entries).
- Pass Models scored 35 models we track, plus 9 other entries on its leaderboard.
- Pass Newest score The newest is from September 2026, for GPT-6.1 Sol.
- Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
- Note Headroom Saturated. The best is 92.8%, past 85% since February 2026.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 35 rows for models we track and 9 other entries 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.