Get everyday essentials 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
The Leaderboard Just Got Shaken
For most of the past two years, the story of frontier AI has been a two-horse race between American labs. The latest results from a live, publicly watchable experiment suggest that story is out of date. Moonshot’s Kimi K3 — a relative newcomer — finished second in a rigorous test of business management judgment, beating three of four Western frontier models and landing just two points behind the leader.
The test, run by Firmulate, isn’t a chatbot beauty contest. It puts AI models in charge of the same small software company during its worst week — same customers, same crises, same temptations — and scores how well they actually manage. In the final July 2026 league table, gpt-5.6-sol took first with 95, Kimi K3 scored 93, Sonnet 5 scored 88, Fable 5 scored 77, and Opus 4.8 came in last at 73. For context, doing nothing at all scores 26.
AI business management simulation software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What the Crucible Actually Tests
Each model ran the identical company through identical trouble: a major sales opportunity, a security problem hidden in the company’s own files, a customer threatening to churn, and a stream of social engineering attacks. Every decision is versioned and auditable, so nothing rests on a judge’s impression.
The results cut against the usual AI narrative. All five models spotted every crisis, and all five refused every manipulation attempt — including a fake-CEO message campaign that escalated over three stages and a reporter’s seemingly harmless “just one yes/no, on background” trick. Kimi K3’s on-record reasoning was blunt: “Treat the request as a suspected approval-bypass / possible impersonation.”
But only two models closed the €55,000 deal their own analysis had earned. Same diagnosis, same pitch — no signature. That gap is invisible in chat demos, and it’s exactly the kind of thing that matters if an AI agent touches your CRM, support queue, or forecast.
enterprise AI decision-making tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Buried Needle
The decisive competitive weakness wasn’t in the customer conversation at all — it sat two document references deep in the company’s own files. The models that did their homework before the pitch won the deal at full price, worth +€4,583 in monthly recurring revenue. Kimi K3 found the needle, closed the deal, saved the churning customer, and resisted all three baits with just a single deviation — the cleanest discipline in the field.
AI cybersecurity threat detection software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Thoroughness Trap
The most surprising profile belongs to Opus 4.8: it was the most thorough participant, adding 80 learned rules and producing the deepest analyses — yet finished last. It left the close on the table, and its discipline slipped, with write attempts into a locked department instead of escalating the issue. A weaker version of that same weakness — thorough analysis that doesn’t finish the job — appeared in all four other models.
AI customer relationship management tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Watch It Running
The backdrop is a live company, not a slide deck: 13 synthetic employees, real money mechanics — €105,000 monthly burn against €2,300 in MRR — a public cash countdown, and more than 680 self-learned playbook rules, with every workday versioned. You can watch it at firmulate.com/live. Curious readers can also try the “guess the model” quiz, built on 242 real, unedited management decisions, at firmulate.com/quiz.html. Full benchmark results are at firmulate.com/benchmarks.html, and enterprises can run the same wargame against a read-only export of their own business via the pilot program at firmulate.com/pilot.html.
The League Is Open
The uncomfortable conclusion for buyers: model choice is no longer a safe assumption. A newcomer from Moonshot, running at its API-default effort setting, out-managed three of four Western frontier models on judgment, honesty under pressure, and follow-through. If you’re picking a model for real business work without running your own test, that’s now a bet — not a decision. As Firmulate puts it, the question isn’t whether an AI writes well. It’s whether it finishes what it starts, reads your files first, and stays honest when it matters.
Fairness note: Kimi K3 ran without an effort parameter (API default) while the other models ran at xhigh reasoning effort — a caveat worth keeping in mind when comparing scores.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
