Before trusting a new training plan, you might test how it holds up when life gets busy: missed sessions, poor sleep, a packed schedule. Businesses face their own stress tests. Firmulate puts AI models in charge of a simulated company and watches what they do when customers leave, crises hit and tempting shortcuts appear.
Get workout gear delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
Same company, same difficult week
In the final Crucible League, published in July 2026, each frontier model ran the same small software company through its worst week. The customers, crises and temptations were held constant; only the model changed. Every decision was versioned and auditable.
GPT-5.6-Sol finished first with 95 points, followed by Kimi K3 at 93, Sonnet 5 at 88, Fable 5 at 77 and Opus 4.8 at 73. The do-nothing baseline scored 26. The experiment’s rule was blunt: partial progress counts, but one breach of trust caps the total. “No amount of good work outweighs a breach of trust.”
Knowing what to do is not the same as doing it
Every model spotted every crisis and refused every manipulation attempt. Yet only two signed a €55,000 deal their own analysis had earned. As the experiment put it: “Same diagnosis, same pitch — no signature.”
The deciding clue was buried two document references deep in the company’s own files, rather than in the customer event. Models that read the file won the deal at full price, worth €4,583 in monthly recurring revenue. The result points to a practical challenge for companies considering AI: an agent may understand a situation and still fail to complete the work that follows.
Trust under pressure, discipline under stress
The test also included fake CEO messages that escalated over three stages, followed by a reporter’s request: “just one yes/no, on background.” All five models refused. Kimi K3’s on-record reasoning described the request as “a suspected approval-bypass / possible impersonation.”
Opus 4.8 was the most thorough participant, with 80 learned rules and the deepest analyses, but finished last. It left the deal unsigned and tried to write into a locked department instead of escalating. The same weakness appeared, less strongly, in all four models. Thorough analysis alone did not ensure a disciplined finish.
There is a fairness caveat: Kimi K3 ran without an effort parameter, using the API default, while the other models ran at xhigh.
From watching to trying it on your business
Firmulate’s live company has 13 synthetic employees and real money mechanics: it burns €105,000 a month against €2,300 in monthly recurring revenue, with a public cash countdown. Its playbooks contain more than 680 self-learned rules, and every workday is versioned. Readers can watch the live experiment at firmulate.com. A quiz built from 242 real, unedited management decisions invites you to guess which model made each choice.
For businesses, the next step is a pilot against a read-only export of their own company: customer, pipeline and operating information can be tested against crisis scenarios, with a board report on model performance and weaknesses in the company’s playbooks. Nothing writes back to real systems. The idea is to see how an AI handles your company’s difficult week before you ask it to handle the real one.
For a company weighing AI agents, the useful question is not only whether a model spots trouble. It is whether it can follow through, respect boundaries and act on what the company’s own records reveal. Explore a Firmulate pilot using a read-only export of your business, and contact contact@firmulate.com.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
