AML Bench (Veris AI)
Agents investigate suspicious activity at a simulated bank. A model judge checks whether the recorded service requests and final case disposition satisfy every requirement.
Veris ranks 14 models on 153 investigations. In 48 cases, the agent must ask a colleague for a missing record.
- Made by
- Veris AI
- Released
- October 2026
- Kind
- Finance
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- Kept private
- Measures
- Capability
- Subject
- Economics and business, Law
- Language
- English
- Interaction
- Agentic
- Environment
- APIs and tools, A simulator
- Refresh
- One fixed set
- Run by
- A host
- In views
- Agents, Robotics, Law and finance
- Strongest model tested
-
Gemini 4 Carbon #3 on the capability index
- Last model added
- November 2026
- Scores refreshed
- November 20th, 2026
GPT-6.1 Sol leads with 70.8%
Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.
- 1 GPT-6.1 Sol 70.8%
- 2
Gemini 4 Carbon 70.4%
- 3 GLM-5.3-Flash 68.6%
- 4
Gemini 4 Argon 68.6%
- 5 GLM-5.3 68.2%
- 6
Gemini 4 Hydrogen 67.5%
- 7 Grok 4.7 67.3%
- 8 Claude Sonnet 5.5 67.3%
- 9 Kimi K3 67.1%
- 10 Claude Opus 5.5 66.4%
- 11
DeepSeek V4.1 Flash 64.5%
- 12 GPT-5.6 Terra 62.3%
- 13
Gemini 3.8 Flash 61%
- 14
Nemotron 3.5 Super VL 56.9%
Source: Official AML Bench (Veris AI) leaderboard
| # | Model | Score ± | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6.1 Sol | 70.8% ±6.9 | Codex | Sep 2026 | Official leaderboard |
| 2 | | 70.4% 63.2 to 77.3 | Gemini CLI | Oct 2026 | Official leaderboard |
| 3 | GLM-5.3-Flash | 68.6% ±6.9 | Pi | Aug 2026 | Official leaderboard |
| 4 | | 68.6% ±7.1 | Gemini CLI | Sep 2026 | Official leaderboard |
| 5 | GLM-5.3 | 68.2% ±7 | Pi | Aug 2026 | Official leaderboard |
| 6 | | 67.5% ±7.1 | Gemini CLI | Nov 2026 | Official leaderboard |
| 7 | Grok 4.7 | 67.3% ±7.1 | Grok CLI | Sep 2026 | Official leaderboard |
| 8 | Claude Sonnet 5.5 | 67.3% ±7.3 | Claude Code | Sep 2026 | Official leaderboard |
| 9 | Kimi K3 | 67.1% ±6.9 | Pi | Jul 2026 | Official leaderboard |
| 10 | Claude Opus 5.5 | 66.4% ±7.3 | Claude Code | Sep 2026 | Official leaderboard |
| 11 | | 64.5% ±7.1 | Pi | Sep 2026 | Official leaderboard |
| 12 | GPT-5.6 Terra | 62.3% ±6.9 | Codex | Jul 2026 | Official leaderboard |
| 13 | | 61% ±6.9 | Gemini CLI | Sep 2026 | Official leaderboard |
| 14 | | 56.9% ±6.6 | Hermes | Oct 2026 | Official leaderboard |
Rankings by part
We read 2 parts from the benchmark’s own leaderboard: Single-turn and Multi-turn. Each ranks models on one part of the test. GLM-5.3-Flash leads Single-turn and Gemini 4 Carbon leads Multi-turn. Arrows show how far a model moved from its overall rank among the models we track.
Single-turn
14 models we track have a score here.
- 1 GLM-5.3-Flash ↑2 91.4%
- 2 GLM-5.3 ↑3 90.8%
- 3 GPT-6.1 Sol ↓2 90.8%
- 4 Kimi K3 ↑5 89.5%
- 5 Grok 4.7 ↑2 89.5%
- 6 Claude Sonnet 5.5 ↑2 89.5%
- 7
Gemini 4 Hydrogen ↓1 87.3%
- 8
Gemini 4 Argon ↓4 86.3%
- 9
DeepSeek V4.1 Flash ↑2 86%
- 10
Gemini 4 Carbon ↓8 86%
Multi-turn
14 models we track have a score here.
- 1
Gemini 4 Carbon ↑1 36.1%
- 2 Claude Opus 5.5 ↑8 30.6%
- 3
Gemini 4 Argon ↑1 29.9%
- 4 GPT-6.1 Sol ↓3 27.1%
- 5
Gemini 4 Hydrogen ↑1 24.3%
- 6 GLM-5.3 ↓1 18.8%
- 7 GLM-5.3-Flash ↓4 18.8%
- 8 Grok 4.7 ↓1 18.8%
- 9 Claude Sonnet 5.5 ↓1 18.8%
- 10 GPT-5.6 Terra ↑2 18.1%
The best score, 70.8%, costs $0.26 a task
Runs on the cost frontier score higher than any run that's cheaper. GLM-5.3-Flash comes within 2.2 points of the best for $0.02 a task, 13 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Official AML Bench (Veris AI) leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-6.1 Sol | Codex | 70.8% | $0.26 | ||
| Gemini CLI | 70.4% | $0.91 | |||
| GLM-5.3-Flash | Pi | 68.6% | $0.02 | ||
| Gemini CLI | 68.6% | $0.53 | |||
| GLM-5.3 | Pi | 68.2% | $0.14 | ||
| Gemini CLI | 67.5% | $0.74 | |||
| Grok 4.7 | Grok CLI | 67.3% | $0.50 | ||
| Claude Sonnet 5.5 | Claude Code | 67.3% | $0.21 | ||
| Kimi K3 | Pi | 67.1% | $0.31 | ||
| Claude Opus 5.5 | Claude Code | 66.4% | $0.44 | ||
| Pi | 64.5% | $0.06 | |||
| GPT-5.6 Terra | Codex | 62.3% | $0.29 | ||
| Gemini CLI | 61% | $0.46 | |||
| Hermes | 56.9% | $0.05 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official AML Bench (Veris AI) leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AML Bench (Veris AI) 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 11 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access The tasks are private.
- Pass Confidence Checked against a primary source.
Scores 5 pass
- Pass Independent runs Scores from the official leaderboard (14 entries).
- Pass Models scored 14 models we track.
- Pass Newest score The newest is from November 2026, for Gemini 4 Hydrogen.
- Pass Frontier models 5 of the top 10 have a score, led by Gemini 4 Carbon (#3).
- Pass Headroom The best is 70.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on veris.ai: 14 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.