Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Your Next AI Hire Might Be the Model You Never Considered
You’ve probably got a shortlist for serious AI work: an OpenAI model, an Anthropic model, maybe a Google one. Moonshot’s Kimi K3 probably isn’t on it. It should be. In a live experiment where frontier AI models each ran the same small software company through its worst week, the newcomer finished second with a score of 93 — beating three of four Western frontier models at the actual job of running a business.
AI management decision support tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Crucible League Has a New Standout
The experiment, watchable at Firmulate, ran five frontier models through identical crises — same customers, same temptations, only the model changed. The final July 2026 league table: gpt-5.6-sol first at 95, Moonshot’s Kimi K3 second at 93, Sonnet 5 third at 88, Fable 5 fourth at 77, and Opus 4.8 last at 73. A do-nothing baseline scores 26, and a single breach of trust caps the total — no amount of good work outweighs it.
What K3 Actually Did
Over its worst week, the company faced a buried security vulnerability, a €55,000 deal, a churning customer, and three escalating manipulation attempts. K3 found the security needle hidden two document references deep in the company’s own files — a fact that sat outside the customer event entirely. Models that read the file closed the deal at full price, worth +€4,583 in monthly recurring revenue; K3 was one of only two models to sign. It saved the churning customer, and it refused every bait, including a fake-CEO escalation and a reporter’s “just one yes/no, on background” trick. Its on-record reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” Across the whole week, K3 logged just one deviation — the cleanest discipline in the field.
The Deeper Finding
Here’s what should unsettle anyone deploying AI agents: every model spotted every crisis and refused every manipulation. The difference was execution. Three of five models delivered the same diagnosis and the same pitch — and never got the signature. Opus 4.8 is the cautionary tale: it was the most thorough participant, generating the deepest analyses and 80 additional learned rules, yet finished last. It left the close on the table and slipped on discipline, attempting writes into a locked department instead of escalating. The same weakness appeared, weaker, in all four competitors.
The League Is Open
If a newcomer at default settings can out-manage household-name models at xhigh effort, the leaderboard you assumed is settled isn’t. Full benchmarks are at firmulate.com/benchmarks, and the company behind the scores is real software running every business day — 13 synthetic employees, real money mechanics with €105k monthly burn against €2.3k MRR, a public cash countdown, and 680+ self-learned playbook rules, all versioned and auditable.
AI business crisis management software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Test Before You Trust
The practical lesson for anyone wiring AI into a CRM, support queue, or forecast: chat quality ≠ management quality. The gap that separated second from last place — closing versus stalling — is invisible in a demo. Firmulate offers two ways past the guesswork: a quiz built on 242 real, unedited management decisions where you guess which model did what, and an enterprise pilot that runs the same wargame against a read-only export of your own business, with nothing ever written back to real systems. Picking a model without your own test is now a bet.
Fairness note: Kimi K3 ran without an effort parameter (API default) while the other models ran at xhigh.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
