Anthropic's Claude Fable 5.2 Scores Below Fable 5.1 on Accounting Work, DualEntry Benchmark Finds

NEW YORK--(BUSINESS WIRE)--Anthropic's newly released Claude Fable 5.2 scored 81.2% on the DualEntry Accounting AI Benchmark, tying for 14th among 23 leading AI models tested.

The result marks a 3.9-percentage-point decline from Claude Fable 5.1, which scored 85.1% on the same benchmark. Fable 5.2 also finished behind Anthropic's own Claude Opus 5.5 at 82.2%.

The model performed strongest in transaction classification and accounting knowledge, but a weak result in accounts receivable left it behind Google's Gemini 4 Argon at 91.1%, Zhipu AI's GLM-5.3 Flash at 88.1% and SpaceXAI's Grok 4.7 at 88.1%.

Key findings:

• Claude Fable 5.2 scored 81.2% overall, tied for 14th of 23
• It declined 3.9 percentage points compared with Claude Fable 5.1 (85.1%)
• Its strongest areas were transaction classification and accounting knowledge, both 100% (13 of 13)
• Its weakest area was accounts receivable at 33.3% (4 of 12), the lowest receivables score on the benchmark
• It scored 86% on easy tasks and 70% on hard tasks
• Its consistency score was 76.2%, down from 80.2% for Fable 5.1
• It costs $1.13 per task, slightly less than Fable 5.1 ($1.18) but about 10 times Gemini 4 Argon ($0.115)
• Gemini 4 Argon scored 9.9 points higher at roughly a tenth of the cost

Claude Fable 5.2 was competitive in journal entries, accounts payable and financial reporting, scoring 84.6% in each, and in bank reconciliation at 83.3%. It struggled more in accounts receivable and month-end close (75%).

Its weakest result came in accounts receivable at 33.3%, compared with Gemini 4 Argon at 91.7% and Claude Fable 5.1 at 58.3%.

Fable 5.2 has posted gains on broader reasoning, coding and agentic benchmarks. Those gains did not carry over to accounting work.

"A newer model isn't automatically a better model for finance," said Benedict Dohmen, co-founder of DualEntry. "Fable 5.2 scores well on general benchmarks but got fewer accounting tasks right than the version it replaces, which is why teams should test models on the actual accounting work they need done before they switch."

The DualEntry Accounting AI Benchmark tests models on 101 synthetic accounting cases across eight areas: transaction classification, journal entries, accounts payable, accounts receivable, bank reconciliation, financial reporting, month-end close, and accounting knowledge.

Each model completes the full benchmark three times from an identical test company with the same tools and instructions. Answers are checked against reference answers written in advance using fixed scoring rules rather than an AI grader. No customer data is used.

Full results for all 23 models, including accuracy, consistency, cost, response time, and performance by accounting area, are available at dualentry.com/benchmark.

About DualEntry

DualEntry is the agentic ERP for mid-market finance teams. It brings the general ledger, accounts payable and receivable, revenue recognition, multi-entity consolidation, and reporting into one platform, with AI agents that take on accounting work from data entry through close. Founded in 2024 and based in New York, DualEntry raised a $90 million Series A in 2025 from Lightspeed Venture Partners, Khosla Ventures, and GV. Learn more at DualEntry.com.