Ferrous helps you size model API commitments, negotiate where you have leverage, and verify savings against actual usage. Independent advice for the people building the product and the people owning the budget.
One purchasing decision. A two-page assessment within two business days after we confirm scope and receive the agreed inputs. No call required to start.
This example shows why utilization belongs in the purchasing decision. A discount has value only in the context of what you will actually use and what the commitment provides.
For your assessment, we distinguish observed facts from estimates and identify any information needed before recommending a change.
Where alternatives require a different model or architecture, they remain candidates until your team validates quality, latency and operating requirements.
Assume a fixed monthly commitment covers usage worth $100,000 at published rates. The discount is 20%, but expected usage is only 70% of that allowance.
| Usage allowance at published rates | $100,000 |
|---|---|
| Fixed bill after 20% discount | $80,000 |
| Expected usage at published rates | $70,000 |
| Cost above usage-based pricing | $10,000 |
| Break-even utilization | 80% |
Decision to examine: reduce the minimum or remain on usage-based pricing unless capacity or other contractual benefits justify the extra cost.
Assumes identical rates before discount, no rollover, no credits, no overage and no additional fees. Actual advice requires your terms and a demand forecast. These are arithmetic assumptions, not market benchmarks or projected savings.
Our focus is B2B AI application teams with recurring model API spend, a clear product workload and a purchasing decision in the next few months.
We are best suited to teams spending roughly $50,000–$250,000 a month on model APIs. Spend outside that range can still be a fit when there is enough actionable value.
We work alongside your finance and engineering leads. Your existing cost tools and financial models become inputs to the decision.
Model APIs first. For GPU-heavy or self-hosted workloads, we confirm fit and technical scope separately.
Fit before fees. If the available evidence does not support a worthwhile engagement, we explain why. A small bill or an already competitive contract may need only a brief review.
Bring us into a client or portfolio company's upcoming purchasing decision. You keep the financial relationship; Ferrous contributes the contract review, purchasing plan and savings evidence.
Start with one company that wants help. Expand to a small portfolio review when the process proves useful.
It can reveal billed rates, credits and questions worth investigating. It cannot establish workload quality, migration feasibility or every contract obligation. The free assessment explains what we can conclude and what additional evidence is needed.
No. Vendor flexibility depends on spend, timing, terms and credible alternatives. Sometimes the better decision is a smaller commitment, a workload change or no change at all. We do not guarantee a discount or a fundraising outcome.
Before an engagement, we agree what the same eligible workload would cost under the baseline, how usage and price changes are treated, and which implementation costs are deducted. Your proposal sets the savings-share percentage, measurement period and payment schedule. We do not treat an automatic public price reduction as a negotiated saving.
The initial assessment starts with agreed billing and contract information. Please do not send credentials, API keys or customer data. If later work needs technical access, the scope and access arrangements are agreed separately with your team.
Not automatically. Any use beyond your engagement requires a separate agreement covering permission, aggregation and confidentiality restrictions. Participation is optional and is not required for the free assessment.
Share your spend range and the decision ahead. We aim to respond within one business day, confirm the scope and agree what information to review.