The moment before you sign is when you have the most leverage and the least time. Here are the levers that are still open on an Anthropic agreement, and how each one moves real money over the term.
The signature is the moment of maximum leverage and minimum time. Before you sign, every term is theoretically open and the vendor wants the deal closed. After you sign, almost nothing moves until renewal, and renewal arrives with the baseline already set against you. So the question that matters in the final stretch is not whether you got a discount, it is which specific levers are still open and which ones actually move money over the life of the agreement. Most buyers focus on the headline rate and leave the levers that matter more untouched. Here is the full set, in the order we work them, and what each one is worth.
The effective rate per token is the obvious lever, but it has to be read by model and after all discounts, not as a single headline number. Opus, Sonnet, and Haiku price very differently, and a deal that looks strong on one can be soft on another. The lever to pull is the effective rate on the models you actually use most, weighted by your real mix. Pushing the rate on a model you barely touch is theater. Pushing it on the model that carries the majority of your spend is where the money is. Before signature, the rate on your dominant models should sit in the discount band your committed volume genuinely commands, and that is the first thing to verify.
The committed spend is the lever that determines almost everything else, because the commit unlocks the discount tier and also sets your exposure. The mistake is to size the commit to current, unoptimized usage or to an optimistic growth story. The right size is set against an optimized baseline and a realistic projection. Optimization first matters enormously here, because routing, caching, and batch can take a large share off your real consumption, and a commit sized to the unoptimized number locks in waste you have already identified. At signature, the commit should be the number you can confidently use after optimization, large enough to reach a good discount band, not so large that you are buying insurance you will forfeit.
One of the most valuable and most overlooked levers is how usage above the commit is priced. In many agreements, overage defaults to an on demand rate that is worse than the committed rate, which means growth past your commit gets punished exactly when it should be rewarded. The lever to pull is to have overage priced at the committed rate, so that exceeding the commit is neutral rather than penal. For a buyer on a rising token curve, this single term can be worth more over the term than a point or two on the headline rate, because the growth is where the volume concentrates. It is also a term vendors will often concede because it costs them little to grant and feels generous to the buyer.
The mirror image of overage is what happens to commitment you do not use. In most agreements, unused committed spend simply disappears at the end of the period, which means any commit set above actual usage is a direct overpayment. The levers here are carryover and rollover, allowing unused commitment to carry into the next period or roll forward rather than vanish. Even partial carryover materially changes the math, because it turns the commit from a use it or lose it bet into something closer to a balance. At signature this is worth pressing on, especially if your usage is uncertain, because it removes the penalty for sizing the commit conservatively.
A rate is only as good as its durability. Without price protection, a strong rate at signature can be eroded by uplift during the term or reset at renewal. The lever is an explicit price lock for the term, and ideally a cap on any uplift at renewal, so the deal you negotiated is the deal you keep. This matters most for multi year agreements, where an unprotected rate can drift well away from where it started. Buyers often celebrate a signature rate and ignore its protection, then discover at renewal that the baseline has quietly moved. Locking the rate and capping the reset is how you keep the win you negotiated.
Term length is a lever in both directions. A longer term can buy a better rate and price protection, which is valuable when your usage is predictable and you want stability. A shorter term preserves flexibility and the ability to renegotiate as the market and your usage change, which is valuable when you are early and uncertain. The lever is to choose the term deliberately as a trade rather than accepting the default the vendor proposes. A long term with strong price protection and good overage treatment can be an excellent deal. A long term without those protections is a trap, because it locks you in without locking the vendor down.
Beyond the headline levers sit several order form details that quietly move money and are easy to sign past. Automatic true forward provisions that step up the commit each period without negotiation. Renewal terms that reset the baseline to list. Definitions of what counts toward the commit that exclude certain usage. Minimums that ratchet up. None of these are dramatic on the page, and all of them can cost real money over the term. The lever is simply to read them with an eye trained on where money moves, and to negotiate the ones that work against you before signature, because after signature they are settled.
Buyers tend to weigh levers by how they feel in the moment rather than by what they are worth over the life of the agreement, and that distorts the priorities. A point off the headline rate feels like a win and gets celebrated, while overage priced at the committed rate feels like a technicality and gets waved through. Yet for a buyer whose usage is growing, the overage term can be worth far more over a multi year agreement than the headline point, because the growth concentrates exactly in the volume that overage governs. The right way to value each lever is to project it across the full term against your real usage curve, not to judge it by how large the concession sounds at signature. Run that exercise and the priorities often invert: the quiet structural terms turn out to carry more money than the loud headline ones.
The levers are not independent dials, they push and pull on each other, and a good negotiator works them as a system. A larger commit unlocks a better discount band but raises your exposure to unused commitment, so it is only worth pulling if you can pair it with carryover or confident utilization. A longer term can buy a better rate but is only safe if it comes with price protection, otherwise it locks you in without locking the vendor down. Overage at the committed rate matters more the more uncertain your growth is, because it removes the penalty for sizing the commit conservatively. Pulling one lever in isolation can quietly worsen your position on another, which is why the sequence and the combination matter as much as any single ask.
Some of the most valuable levers cost the vendor very little to grant, which makes them the easiest wins to secure if you know to ask. Overage at the committed rate is often one of these, because it feels generous to the buyer while costing the vendor little in expectation. A renewal uplift cap can be another, since it constrains a future the vendor was not certain to realize anyway. Partial carryover of unused commitment can be a third. A prepared buyer presses on these precisely because the vendor's resistance is low relative to the buyer's gain, and securing them does not spend the goodwill needed for the harder asks like the headline rate. Knowing which concessions are cheap to the other side is a large part of negotiating efficiently.
The levers all work better when the buyer arrives with an optimized, well measured consumption story rather than an unmanaged one. A vendor account team can see the difference between a buyer who knows their model mix, cache hit rate, batch share, and real baseline, and one who is signing against a vague sense of usage. The informed buyer gets better terms not only because they ask for the right things but because the vendor recognizes a sophisticated counterparty who will hold them to the deal and who has alternatives. Optimization done before the negotiation is therefore not just a way to shrink the commit, it is a way to change how seriously the vendor takes every lever you reach for.
Pulling the levers in the right sequence matters as much as pulling them. Optimize the spend first, so you know the real baseline. Size the commit to that optimized number and the discount band it should reach. Secure overage at the committed rate and the best treatment of unused commitment you can get, because those govern your exposure in both directions. Lock the rate and cap the renewal reset to protect the win. Choose the term deliberately as a trade for protection. Then read the fine print for the quiet costs. Worked in that order, the levers compound, and the deal that results is sized to reality, protected over time, and free of the penalties that catch unprepared buyers.
These levers are the negotiable surface of Anthropic enterprise pricing. For the underlying rate bands, discount tiers, and commit mechanics, read the pillar guide on Anthropic and Claude pricing in 2026, and book a strategy call before you sign so we can walk every lever on your specific draft.
Book a strategy call before you sign. We will walk every lever on your draft and tell you which ones are worth pulling.
Book a Strategy CallWeekly intelligence on Anthropic pricing moves and the buyer side counters that work.