Benchmark Almanac

Ramp Stack Accounting Benchmark

Ramp Stack 65.8% tops its leaderboard Claude Opus 4.6 62.1% leads the models we track
Not in model cards

Agents perform month-end accounting work across eight simulated businesses. Ramp measures the fraction of grading criteria each agent satisfies.

Ramp compares its accounting agent with frontier models across 237 tasks.

Made by
Ramp Labs
Released
June 2026
Kind
Finance
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
Kept private
Measures
Capability
Language
English
In and out
Text, Tables → Text, Actions
Interaction
Agentic
Environment
A simulator
Items from
Generated
Refresh
One fixed set
Run by
A host
Occupations
Business and financial operations
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
May 2026
Scores refreshed
November 20th, 2026
7
Entries on its own leaderboard, 6 models we track

Ramp Stack tops the leaderboard with 65.8% Claude Opus 4.6 leads with 62.1% among models released by May 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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. The best model we track is Claude Opus 4.6, at 62.1%. This view shows only models we track, each at its best.

The newest model with a score here came out in May 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Ramp Stack Skills and memory 65.8%
  2. 2 Claude Opus 4.6 No skills or memory 62.1%
  3. 3 GPT-5.5 No skills or memory 61.7%
  4. 4 Claude Opus 4.7 No skills or memory 60.7%
  5. 5 Gemini 3.1 Pro No skills or memory 54.3%
  6. 6 Gemini 3.5 Flash No skills or memory 45.7%
  7. 7 GPT-5.4 No skills or memory 42.7%

Source: Official Ramp Stack Accounting Benchmark leaderboard

#EntryScore ±SettingDateSource
1 Ramp Stack 65.8% 62.5 to 69.4 Skills and memory Official leaderboard
2 Claude Opus 4.6 62.1% ±4.8 No skills or memory Feb 2026 Official leaderboard
3 GPT-5.5 61.7% 56.5 to 66.1 No skills or memory Apr 2026 Official leaderboard
4 Claude Opus 4.7 60.7% 53.2 to 67.7 No skills or memory Apr 2026 Official leaderboard
5 Gemini 3.1 Pro 54.3% ±7.4 No skills or memory Feb 2026 Official leaderboard
6 Gemini 3.5 Flash 45.7% ±4.5 No skills or memory May 2026 Official leaderboard
7 GPT-5.4 42.7% 38.3 to 47.4 No skills or memory Mar 2026 Official leaderboard
#ModelScore ±SettingDateSource
1 Claude Opus 4.6 62.1% ±4.8 No skills or memory Feb 2026 Official leaderboard
2 GPT-5.5 61.7% 56.5 to 66.1 No skills or memory Apr 2026 Official leaderboard
3 Claude Opus 4.7 60.7% 53.2 to 67.7 No skills or memory Apr 2026 Official leaderboard
4 Gemini 3.1 Pro 54.3% ±7.4 No skills or memory Feb 2026 Official leaderboard
5 Gemini 3.5 Flash 45.7% ±4.5 No skills or memory May 2026 Official leaderboard
6 GPT-5.4 42.7% 38.3 to 47.4 No skills or memory Mar 2026 Official leaderboard

Rankings by part

We read 4 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Variance Analysis, Reconciliation and Data Entry. Claude Opus 4.7 leads 2 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Variance Analysis

6 models we track have a score here.

  1. 1 GPT-5.5 ↑1 69.9%
  2. 2 Claude Opus 4.6 ↓1 66.5%
  3. 3 Claude Opus 4.7 65.1%
  4. 4 Gemini 3.1 Pro 63.2%
  5. 5 Gemini 3.5 Flash 53.4%
  6. 6 GPT-5.4 47.1%

Accuracy versus runtime

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

40% 50% 60% 70% 80% 05001,0001,500 Runtime per task (s), lower is better Mean criteria accuracy Ramp Stack (Skills and memory)GPT-5.4 (No skills or memory) 40% 50% 60% 70% 80% 05001,0001,500 Runtime per task (s), lower is better Mean criteria accuracy Ramp Stack (Skills and memory)GPT-5.4 (No skills or memory)
ModelSettingMean criteria accuracyRuntime per task
Ramp Stack Skills and memory 65.8% 364 s
Claude Opus 4.6 No skills or memory 62.1% 644 s
GPT-5.5 No skills or memory 61.7% 609 s
Claude Opus 4.7 No skills or memory 60.7% 605 s
Gemini 3.1 Pro No skills or memory 54.3% 662 s
Gemini 3.5 Flash No skills or memory 45.7% 1,269 s
GPT-5.4 No skills or memory 42.7% 166 s

Accuracy versus runtime

The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

40% 50% 60% 70% 80% 05001,0001,500 Runtime per task (s), lower is better Mean criteria accuracy Claude Opus 4.6 (No skills or memory)GPT-5.5 (No skills or memory)Claude Opus 4.7 (No skills or memory)GPT-5.4 (No skills or memory) 40% 50% 60% 70% 80% 05001,0001,500 Runtime per task (s), lower is better Mean criteria accuracy Claude Opus 4.6 (No skills or memory)GPT-5.5 (No skills or memory)Claude Opus 4.7 (No skills or memory)GPT-5.4 (No skills or memory)
ModelSettingMean criteria accuracyRuntime per task
Claude Opus 4.6 No skills or memory 62.1% 644 s
GPT-5.5 No skills or memory 61.7% 609 s
Claude Opus 4.7 No skills or memory 60.7% 605 s
Gemini 3.1 Pro No skills or memory 54.3% 662 s
Gemini 3.5 Flash No skills or memory 45.7% 1,269 s
GPT-5.4 No skills or memory 42.7% 166 s

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard Not tested since Jun 2026

Source: Official Ramp Stack Accounting Benchmark leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 4.6 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Claude Opus 4.6

Source: Official Ramp Stack Accounting Benchmark 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

All 637 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 Newest score check. The newest came 6 months ago, in May 2026, for Gemini 3.5 Flash. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (7 entries).
  • Pass Models scored 6 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 6 months ago, in May 2026, for Gemini 3.5 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.5 (#21).
  • Pass Headroom The best is 62.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on builders.ramp.com: 6 rows for models we track and 1 other entry 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.

How we check