Run 2026-09-05 · one company's GTM stack · redacted

A benchmark built from the work, not from a leaderboard.

Public model leaderboards test abstracted skills. None of them measure operations work: extracts with blank columns, joins that lie, renewal dates that disagree between systems. BonuslyBench closes that gap with 40 tasks drawn from a real revenue organization — then grades every answer with deterministic code.

01 · Why

The question the exercise set out to answer

When language models are pointed at the actual daily work of revenue operations — reconciling a rollforward, auditing stage hygiene, building a weighted forecast — what actually happens? Not on vendor benchmarks. On real, messy data, under a grading standard that cannot be charmed.

Every GTM organization is being asked to believe that pointing an LLM at a CRM produces reliable analysis. This run replaces belief with measurement: 103 models, 40 tasks, one deterministic rubric.

02 · How

How the benchmark works

40 real tasks, 8 categories

CRM and data hygiene, deal intelligence, rep performance, reporting and analytics, customer success, marketing, executive communication, ops maintenance. Each task hands the model real CSV extracts and asks for the deliverable an operator produces weekly.

Traps planted from production

Owner columns blank on most rows. Renewal dates that disagree across systems. Call mappings keyed to dead deal ids. "LONG(n)" placeholders. A correct answer navigates the trap — or says the data cannot support the claim.

Deterministic grading, no judge

Expected answers were computed programmatically from the same data the models saw. Responses are graded by fixed code: numeric tolerance bands, regex presence checks, alias whitelists. No LLM graded any response. The same rubric scored all 103 models.

Anonymized by design

Company identities are aliased; personal names were replaced with consistent pseudonyms across every test, response, and artifact — verified to be score-neutral (zero grading drift across all 4,576 answers).

03 · Findings

What the run found

Verdict

Open models matched the frontier on this work — and 17 of them cost under $2 for the entire suite.

The top of the leaderboard is open weights. The best frontier entry ties with an open model that costs a fraction as much. Premium pricing bought speed and polish here, not measurable accuracy on operational tasks.

No model aced the suite

The best score was 36 of 40 perfect tasks. The spread opened on tasks that mix arithmetic with judgment: which deals to exclude, what stale means, when to say the data cannot answer.

Fabrication was rare and catchable

20 of 103 models invented an entity at least once. A deterministic alias whitelist caught every instance. The failure mode is real, and it is guardrail-able.

The grader needed grading

Five scoring defects initially made six tasks "impossible" for every model. A verification pass read real answers against real truth and fixed the rubric — the audit protocol is itself a finding.

Speed is its own axis

Median response time ranged from under 20 seconds to over 40 minutes per task. The pool holds fast-and-accurate and slow-and-accurate profiles at similar cost — a real choice for embedded workflows.

04 · Basis

Scope, caveats, and how to read the numbers

WhatDetail
Run2026-09-05, one execution of the full 40-task set
Models103 · 96 with full 40-task coverage · 7 partial (timeout-tier, disclosed)
Reference modelanthropic/claude-sonnet-5
GradingDeterministic Python against frozen truth files · no LLM judge
DataOne company's GTM stack, redacted · company aliases · personal pseudonyms
ComparabilityScores are comparable within this run and rubric version only

Basis: 40 tasks, deterministic rubric, one run. A fabrication flags the response; disclosed partials are never scored on unrun tasks.

Scores describe operational skill on this distribution of work, not general capability. Treat the rankings as a strong prior and verify against your own data before committing spend — the test specification is included so that's a week of work, not a mystery.