Gurobean is a free, educational tool built to make the mechanics of optimization tangible through the fun, real-world scenario of running a coffee shop against shifting constraints. It reflects our commitment to open education: building quality resources that break down barriers to learning, not creating new ones. Like the earlier Burrito game, Gurobean will keep evolving as we hear from the academic community putting it to use.
Like any live game, Gurobean keeps evolving. v2.1 brings a set of fixes and enhancements that make matches run smoother and scenarios feel more true to life.
Fixes
More accurate "optimal" solutions. A bug in our post-processing code occasionally displayed an inaccurate optimal solution, letting a player's score edge out Gurobi's when it shouldn't have. The Gurobi solver itself was never the issue—the fix lives in how the game rounded fractional solutions to integers on the solution screen. Now the label matches the math.

Before the fix, a bug could show the wrong "optimal" solution -- in this case, the optimal number of cups of cold coffee to brew is actually 9, not 8 as shown.

After the fix, the correct optimal solution is displayed. (In this screenshot, the player found the optimal solution, too.)
More realistic Championship Mode scenarios. Some past matchups constrained players into strange corners, like a shop that could brew only one hot cup per hour. New generation rules produce scenarios that better reflect the range of decisions a real coffee shop owner would face.
Constraint Monitor, rendering correctly. The chart tracking cups hot against cups cold sometimes scaled its axes incorrectly, flattening the feasible region into an uninformative shape. It now renders at the right scale, so the trade-off space is visible at a glance.
Enhancements
More data visible by default. Players told us they wanted to see average customer arrivals per hour and the average number of customers a barista can serve without digging through an expanded table. Both numbers are now shown up front.
Clearer scoring language. We used to describe results as a percentage "less than the optimal expected value." That wasn't quite accurate—scores are compared against simulated profit, not expected value. The message now reads "x% less than Gurobi's score."
Smoother heatmap. The flags marking each simulated day used to judder slightly as they moved up the center heatmap. This is now fixed.
Sharper slider design. Small visual refinements to the slider tracks used to set your inputs each round.
Play the update
Take on the new scenarios and see if you can match Gurobi's solution.
Gurobean is free to play—access the game here. We'll keep shipping updates like this one as Gurobean grows.

