Schrodinger does not publish a price for its software, and neither do most of its competitors. The directory listing on Capterra reads contact vendor for pricing, with no free trial and no free version. The academic site license page on Schrodinger's own site lists what is included and then asks you to book a consultation. So the honest answer to what a Schrodinger license costs is that the number is assembled per customer, and the only figures in public circulation are secondhand.
That is an unsatisfying answer, so this piece does the next most useful thing: it explains how the licensing model works, so you can read a quote when it arrives, tells you what is genuinely free, and gives you the published ranges that do exist elsewhere in this market to budget against. It is written for a small antibacterial team deciding whether a full computational chemistry suite is the right line item at all.
How Schrodinger licensing actually works
The mechanism matters more than any single number, because it is what makes two quotes for the same software differ by an order of magnitude. Schrodinger sells floating licenses plus tokens. Tokens are a shared, interchangeable pool: they are drawn down when someone starts a job, and the same pool feeds different applications, so Glide, Jaguar and MacroModel all pull from the same allowance rather than each needing their own seat. University computing centers that host the software describe it this way in their own documentation, which is the most reliable public description of the model available.
Two consequences follow. First, your token count is effectively a concurrency limit, not an annual budget of calculations in the way a credit system would be: it caps how many jobs can run at once. A team that wants to launch four docking runs in parallel needs more tokens than a team that runs one at a time, even if they do the same total work over a year. Second, because tokens are interchangeable across applications, breadth is cheaper to add than depth. Widening the pool costs less than you might expect; letting six people all run heavy jobs simultaneously costs more.
The other half of the quote is which suites you license at all. The academic site license page lists small molecule drug discovery, biologics discovery and materials science as separate suites, alongside active learning technologies, PyMOL 3, seats on the certification courses and teaching materials. A commercial quote is built the same way, from a selection rather than a catalog number, which is exactly why the vendor will not put a figure on a web page.
What you can get without paying
There is a free tier and it is worth knowing its boundary precisely, because the boundary is the reason people end up buying. Maestro Viewer is available to academic users as a free download, with the license installed automatically. It gives you the Maestro interface and lets you visualize and manipulate 3D structures. What it does not include is molecular property prediction or the ability to run simulations.
In other words, the free version is a viewer, not an engine. You can look at a structure and you can learn the interface, which genuinely matters if you are deciding whether your team will adopt the environment. You cannot answer a scientific question with it. Anyone telling you they run their discovery work on the free Schrodinger download is describing something else.
The number that circulates, and why not to budget on it
Search for Schrodinger pricing and you will repeatedly meet a figure of $7,500 a year for an academic Premium package with 30 tokens, sometimes with a $10,000 variant at 40 tokens attached to it. Before that lands in a budget, three things are worth knowing. It is academic pricing, so it says nothing about a commercial quote. It is described in the sources that carry it as historical. And it does not appear on Schrodinger's own pages or on the live directory listing, which today shows only contact vendor for pricing.
This is a specific and repeatable failure mode when researching software prices. Aggregator pages and AI summaries keep quoting figures long after the underlying listing has changed, and the citation chain gets shorter every time the number is repeated until it reads like a fact. The same thing happens with compound catalog sizes, where search snippets still carry an older MedChemExpress count than the vendor's live page shows. If a number is going into a budget, open the primary source and read it there.
What published ranges exist in this market
Since Schrodinger publishes nothing, the usable anchors come from the one vendor in this category that does put ranges in writing. Optibrium states these figures in its own knowledge base, which makes them vendor statements rather than estimates, and they bracket the category well enough to budget against.
| What is being priced | Published figure | What it covers |
|---|---|---|
| Small molecule design platform (StarDrop) | From about $10,000 a year, into six figures | Per user, site or floating licensing, priced by module |
| 3D molecular modelling software, as a category | $50,000 to $1,000,000 a year | License only, before compute, support, training and maintenance |
| BioPharmics 3D modelling subscription | $100,000 to $250,000 a year | Annual subscription, scaled by organization size |
| AI discovery platform (Cerella) | $100,000 and up | Annual subscription, scaled to data and deployment |
Schrodinger is a larger and broader vendor than Optibrium, so treat the upper half of that table as the more relevant comparison for a multi-module commercial suite. The point of the table is not to put a price on a competitor's product, which would be a guess. It is to establish that the category starts in five figures for a focused platform and reaches six or seven for a full modelling environment, so a team that has budgeted four figures for this is not in the market yet and should know that before the first call.
What moves a quote in this category
Five variables do almost all of the work, and knowing them lets you compare offers that arrive in different shapes. Seats and concurrency come first, because they drive setup, training, support load and compute together. Module selection is second, since these suites are sold in pieces and the figure a colleague quotes you may cover a different set than you need. Deployment is third: cloud hosting is usually a modest addition, while on premise brings nodes, a scheduler and someone to maintain them.
Contract term is fourth, and it is the one most often mispriced against reality. A multi year commitment buys a discount that is excellent value if the program is funded that long and a liability if the science pivots, which for a grant funded antibacterial program it frequently does. Validation is fifth and is the item most often discovered after the budget closed. A deployment that has to satisfy GxP or 21 CFR Part 11 is a different commercial product even when it is the same software, and the obligations that come with a validated environment carry their own documentation and change control costs on top of the license.
Do you need a full suite for an antibacterial program?
Only if you are doing structure based design across several target classes. A suite of this kind buys breadth: docking, dynamics, free energy, quantum mechanics and cheminformatics in one environment, driven by someone whose job is to drive it. If you have that person and that scope, the price is defensible and the alternatives are worse.
A large number of antibacterial programs do not have that scope. The recurring question is narrower and it repeats: of these compounds, which ones are worth a plate, against which organisms, and why. That is a prioritization question rather than a modelling question, and it is answerable from structure plus published activity for the closest characterized analogs. Priced as a monthly subscription it is a different order of magnitude: $17,880 a year at the top plan here, about $0.15 per compound against a full panel, against a published range that starts at $50,000 a year for a modelling environment. Those are not competing products, and the comparison only matters because teams routinely reach for the expensive one to answer the cheap question.
The full disclosure picture across the category, vendor by vendor with the source for each row, sits on the drug discovery software cost page. If the specific job is ranking structures before they reach a plate, that is compound library screening software, and the single compound case is MIC prediction software.
How much is a Schrodinger software license?
Schrodinger does not publish commercial pricing, and the current directory listing shows contact vendor for pricing with no free trial and no free version. The quote is built from which suites you license, how many floating tokens you need for concurrency, seat count, contract term and deployment. Academic site licenses are priced separately and also by consultation.
Is Schrodinger software free for academics?
Partly. Maestro Viewer is free for academic users and installs its own license, but it only visualizes and manipulates 3D structures. It does not run simulations or molecular property predictions. Full academic access goes through a site license that Schrodinger prices by consultation, and many universities buy campus wide rather than per lab.
What are tokens in a Schrodinger license?
Tokens are a shared pool of floating license units drawn down when a job starts, and they are interchangeable across applications, so Glide, Jaguar and MacroModel all consume from the same allowance. In practice your token count sets how many jobs can run at once. More parallel users, or heavier simultaneous jobs, means more tokens rather than a different product.
What to get in writing before you sign
Ask for the quote itemized by suite, by module and by token count, so you can see what you would be dropping if the budget moves. Ask what compute is included and what it costs if you exceed it. Ask what year two costs, explicitly, because renewal is where multi year discounts stop protecting you. Ask whether the term can be aligned to a grant cycle. Ask what the validation package costs if there is any chance you will need it.
Finally, ask what happens to your data and your trained models if you stop paying. That question gets skipped and it is the one with the longest tail. The equivalent question is worth asking of any narrower tool too, including this one, and the way the narrow category bills is worked through in the antibacterial screening software pricing write up, where the cost per compound arithmetic is set against what the same compounds cost in plates.
Vendor pricing and packaging move. Every figure here was read at its source at the time of writing, and the figures attributed to Optibrium are that vendor's own published statements about its own products rather than estimates about anyone else's. Confirm against the vendor before a budget depends on it.