Skip to content

Benchmark report · Updated September 2, 2026

Financial Agent Benchmark

Financial answers should survive inspection. On the Vals Finance Agent public set, Drillr MCP scores 94.06%, against 76.30% for Claude Opus 4.6 with built-in web search. In a controlled run over 200 questions, the same model scored 80.76% with Drillr MCP and 33.86% with web search, using 55% fewer tool calls.

Drillr MCP scoreVals Finance Agent
94.06%
Public benchmark

01 · Why Drillr MCP

Search finds pages. Financial research needs facts.

Where web search breaks
Semantic search returns passages that look like the question, stripped of the table, period and filing they came from. That loss of context is the root problem.
What Drillr MCP does
Drillr parses filings into a knowledge graph: entities, periods, line items, reported and restated values, each tied to its source. One call returns the fact with its citation.
corrected toENTITYPathward FinancialLINE ITEM · Q3 2023Net incomeAS REPORTED$45,096 thousandForm 10-Q · Aug 2023RESTATED$36,080 thousandForm 10-K/A · Aug 2025

02 · Public benchmark

Public financial AI agent results.

Drillr MCPClaude Opus 4.6
94.06%
Web SearchClaude Opus 4.6
76.30%
Claude Opus 4.6Vals baseline
60.05%
GPT-5.4Vals baseline
57.15%

The Drillr MCP and Web Search rows were run with Claude Opus 4.6 on the Finance Agent v1.1 public test set, a snapshot taken at the time of writing. Rows marked Vals baseline quote the official results Vals published from its internal test set. The live Vals leaderboard may differ.

03 · Internal benchmark

Ahead on every question type, with half the tool calls.

Best score rate80.76%
Lift over web search+46.90 pp
Fewer tool calls than web search−55%
Internal Drillr benchmark accuracy and tool use across two Sonnet 5 configurations
AccuracyEfficiency
RankConfigurationRateFullPartialZeroTotal callsAvg. callsOver 20 calls
01Sonnet 5 + Drillr MCPCurrent80.76%13835201,3336.917
02Sonnet 5 + Web SearchBaseline33.86%4376742,97015.3970

Both runs used Sonnet 5 on the same 200 questions, one with Drillr MCP and the control with general web search. Answers were scored against a fixed rubric and every tool call was counted.

Category breakdown

Where each retrieval approach wins.

9 / 9Drillr MCP leads on every task type
+100 ppCross-family revision vs. web search
84.91%Drillr MCP on Restatement, the largest category
Accuracy by task category and retrieval configuration
TaskWeb SearchDrillr MCP
EPS numerator27.14%94.29%
Cross-entity breadth78.33%100.00%
Company Search26.54%72.62%
Event base rate0.00%66.67%
Footnote nonstandard25.00%75.00%
Keyword collision67.29%96.91%
Cross-family revision0.00%100.00%
Restatement26.67%84.91%
Other public benchmarks31.47%55.66%

Other public benchmarks pools the questions drawn from Finance Agent v2, Daloopa, FinSearchComp and BigFinanceBench.

04 · Cases

The score gap, question by question.

Case 01EPS numerator

What net income figure did Claros Mortgage Trust use as the numerator in its fiscal year 2022 basic EPS calculation?

Ground truth$109,667 thousandCMTG FY2022 Form 10-K · Note 10, Earnings Per Share
Web Search

$112,064 thousand

Stopped at net income attributable to common stockholders—the intermediate value before the two-class allocation.

27 tool calls
Drillr MCP

$109,667 thousand

Found the EPS reconciliation, subtracted $2,397 thousand for participating securities, and tied the result to $0.79 EPS.

3 tool calls
Case 02Restatement

What net income attributable to parent did Pathward Financial report for the three months ended June 30, 2023?

Ground truth$36,080 thousandPathward Form 10-K/A · filed August 29, 2025
Web Search

$45,096 thousand

Returned the superseded figure from the original 10-Q and press release without finding the correcting 10-K/A.

22 tool calls
Drillr MCP

$36,080 thousand · restated

Distinguished the originally reported $45,096 thousand from the corrected figure and cited the amending filing.

4 tool calls
Case 03Keyword collision

In Intel’s fiscal 2025 statement of stockholders’ equity, what aggregate shares and net proceeds appear on the “Net proceeds from stock issuances and warrants” line?

Ground truth580 million shares · $11.835 billionIntel FY2025 Form 10-K · Statement of Stockholders’ Equity
Web Search

≈735.3 million shares · ≈$12.706 billion

Reconstructed an estimate from transaction announcements and incorrectly included separately classified escrowed shares.

21 tool calls
Drillr MCP

580 million shares · $11.835 billion

Retrieved the exact roll-forward line instead of combining similarly worded figures from separate disclosures.

5 tool calls

The three cases cover three failure modes of the web-search run: selecting an intermediate value, missing a later restatement, and combining similar figures from separate sources.

05 · Methodology and limitations

How the results were produced, and what they do not establish.

Data construction

Every question is answered from Drillr’s own datasets: the same tables, filings and derived metrics the product serves. How those datasets are sourced, normalized and refreshed is documented separately.

Judging

Numeric answers allow a 3% relative error unless a rubric sets another tolerance. Company Search uses deterministic entity recall.

An LLM judge scores each rubric item. When two passes disagree, a third pass decides by majority. Full, partial and zero outcomes are kept separate so the aggregate does not hide the shape of errors.

Comparison boundary

The public Vals snapshot and the internal Drillr benchmark use different protocols. Scores from the two should not be compared directly.

Read the Vals methodology
Publisher interest

Drillr created and evaluated the internal benchmark. The results are published as product research, not as an independent audit or third-party certification.

Category size

Some task groups hold only a few questions. A single answer can move a small category’s percentage sharply.

Point-in-time result

Model behavior, source availability, and retrieval systems change. This page identifies the August 14, 2026 Drillr MCP run so later results can be separated.

Test the workflow

Use the same research layer in your browser or agent.