Qwen4-Max Falls Short of Top Accounting AI Rankings

John Iwuozor
Content Lead, Finance and Research
John Iwuozor
Content Lead, Finance and Research

John Iwuozor leads content, finance, and research at DualEntry. He writes for the people who actually close the books: controllers running month-end, CFOs weighing an ERP migration, accountants buried in reconciliations. Before DualEntry he spent five years covering B2B SaaS and fintech for Forbes Advisor, Ramp, and Infosecurity Magazine. He builds every piece on primary data, including benchmarks, practitioner interviews, and product testing, rather than recycled advice. He is also an avid chess player.

Learn about our editorial policies.
Last updated
November 16, 2026
Reviewed by

Learn about our editorial policies.
Contents
More

Subscribe to the
DualEntry Newsletter

Get fresh Al finance insights, reports and more delivered straight to your inbox.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Summarize this article
TL;DR

Alibaba has unveiled Qwen4-Max, bringing a new flagship model to its growing lineup of AI systems.

‍

We tested Qwen4-Max on the DualEntry Accounting AI Benchmark, which evaluates leading AI models on 101 accounting tasks across eight areas, including journal entries, accounts payable, bank reconciliation, financial reporting, and month-end close.

‍

Qwen4-Max scored 83.2% accuracy, ranking joint ninth on our leaderboard.

‍

It finished behind eight models, including Gemini 4 Argon (91.1%), GLM-5.3 Flash (88.1%), and Kimi K3.1 Pro (86.1%).

Key findings

  • Qwen4-Max scored 83.2% overall accuracy, passing 84 of 101 accounting tasks.
  • It achieved 100% accuracy in transaction classification, financial reporting, and accounting knowledge.
  • Journal entries followed at 84.6%, while bank reconciliation reached 83.3%.
  • Month-end close was its weakest category at 50%, followed by accounts payable at 69.2%.
  • Consistency reached 76.2%, with 77 of 101 tasks passed in all three runs.
  • Qwen4-Max cost approximately $0.025 per task, or $7.64 for the complete three-run benchmark.

Where Qwen4-Max performed best

Qwen4-Max's strongest results came in transaction classification, financial reporting, and accounting knowledge, where it achieved 100% accuracy across all three categories.

‍

Across the eight accounting areas, it scored:

‍

  • Transaction classification: 100% (13 of 13)
  • Journal entries: 84.6% (11 of 13)
  • Accounts payable: 69.2% (9 of 13)
  • Accounts receivable: 75% (9 of 12)
  • Bank reconciliation: 83.3% (10 of 12)
  • Financial reporting: 100% (13 of 13)
  • Month-end close: 50% (6 of 12)
  • Accounting knowledge: 100% (13 of 13)

‍

The perfect scores in classification and financial reporting suggest Qwen4-Max handles structured accounting questions and calculations well. It also performed reasonably well on journal entries and bank reconciliation, passing 21 of the 25 tasks across those two categories.

‍

However, its performance declined when tasks involved more complex accounting adjustments and operational workflows.

Month-end close was Qwen4-Max's biggest weakness

Qwen4-Max passed only 50% of month-end close tasks.

‍

The model struggled with several adjusting-entry scenarios, including prepaid rent amortization, debt reclassification, and allowance for doubtful accounts adjustments.

‍

These are important accounting tasks. Month-end close often involves recognizing expenses in the correct period, adjusting balances, and ensuring that financial statements reflect the company's position accurately.

‍

Accounts payable was another weak area, with nine of 13 tasks passed (69.2%). Accounts receivable followed at 75%, with nine of 12 tasks passed.

How consistent was Qwen4-Max?

The benchmark runs each model through the same 101 tasks three times. Consistency measures the percentage of tasks passed in all three attempts.

‍

Qwen4-Max achieved 76.2% consistency, passing 77 tasks in all three runs. This matters for accounting work because a model that handles a task correctly once may not produce the same result when asked to repeat it.

Qwen4-Max was relatively inexpensive to run

Cost was one of Qwen4-Max's more encouraging results.

The model averaged approximately $0.025 per task, with a total estimated cost of $7.64 for all 303 benchmark attempts. That puts it among the less expensive models on our accounting leaderboard, particularly considering its 83.2% overall accuracy.

‍

However, lower costs don't necessarily translate into better accounting outcomes. A task that requires repeated attempts or additional review can become more expensive in practice.

‍

Qwen4-Max's results suggest it can be economical for accounting work it handles reliably, but its weaker performance in month-end close and payables remains a limitation.

Where Qwen4-Max stands

Qwen4-Max currently ranks joint ninth on the DualEntry Accounting AI Benchmark with 83.2% accuracy.

‍

Its perfect scores in transaction classification, financial reporting, and accounting knowledge demonstrate strong capabilities in those areas. But its 50% month-end close result, weaker payables performance, and 76.2% consistency show that it still struggles with some accounting workflows.

‍

The model's low cost is promising. Whether that translates into practical value will depend on the work it's asked to perform and how reliably it can complete it.

‍

See the full DualEntry Accounting AI Benchmark to compare the latest models across accuracy, consistency, cost, speed, and accounting task area.

‍

See the full power of DualEntry in 30 minutes

Go live in 24 hours

By clicking "Schedule Demo" you agree to the use of your data in accordance with DualEntry's Privacy Notice, including for marketing purposes.