Add each food with its amount and the minute you eat it (0–120 min). Carbohydrate foods take grams of carbohydrate; other foods (protein, fruit, vegetables) take grams as eaten; water takes millilitres. The model empties the stomach (liquid first, solids after a lag, both under a caloric cap), then absorbs glucose and fructose through separate transporters. Rates are shown in g/h.
Every row shares the total-energy cap with whatever is still in the stomach: the calories of other foods displace carbohydrate calories one for one. Protein and fat do not lower the total cap; they slow the carbohydrate's share (CCK, up to −10% for protein, −20% for fat). Extra water adds no calories, leaves with a 13-min half-time, and gives a dilution bonus that decays with it.
Model. Liquids empty first-order with no lag; T½ is 25 min alone and lengthens by 0.05 min per kcal of solid present in the stomach (35 min with 200 kcal, up to 50 min), so a liquid drunk on top of a solid meal empties slower (Collins 1991/2002). Concentrated liquids are governed by the caloric cap. Solids and semisolids wait a lag (bread 20 min; rice-flour paste and grits 8–10 min; thick oat porridge 25–30 min), then empty under the cap. The cap is a total-energy limit of 2.5 kcal/min (Hunt 1985) that does not depend on macronutrient; dilute liquids (at most 0.34 kcal/mL, an 8% carbohydrate drink) raise it by up to +0.4 kcal/min in proportion to their share of the stomach's calories (Vist & Maughan 1995; Brener 1983). While a caloric liquid is in the stomach there is an early uncontrolled phase before duodenal braking is established: liquid delivery runs at up to 1.55× (≤300 mL) to 1.8× (≥800 mL) the steady cap (Hunt 1985: ~4.5 → 2.6 kcal/min, ≈1.7×; Kwiatek 2009 for the volume dependence). The brake follows recent caloric delivery to the duodenum with a 15-min dead time and then an 8-min lag (faster for energy-dense liquids; off τ 20 min), so most of the overshoot is gone by about 40 min and a liquid drunk while a meal is already braking gets only ~1.1× (Berry 2003; Schirra 2000). The overshoot is granted to liquid demand only; solids ramp up through their lag as a sigmoid (width 6 min) instead of switching on (Siegel 1988). A hard run removes the overshoot and depresses emptying 18% at the finish, recovering with τ 25 min. Ginger rises for about 90 min after the dose and fades with a 2-h half-life; walking, posture and sex scale both the demand and the cap; the accelerators add (not multiply) and are capped at +40%. Dehydration, the hard-run slowdown, water and fiber act on the cap only: dehydration slows emptying 5% per percent of body mass lost beyond 2%, floored at −15%, and eases as fluid (water and liquid foods) is drunk (Rehrer 1990; Ryan 1998). Gelatinized starch reaches the absorption pool 10 min after entering the duodenum, free glucose 2 min. Glucose transport is capped at 1.2 g/min and fructose at 0.5 g/min, both × (kg/70)0.5 as a body-surface-area proxy. The dashed line is maximal glycogen synthesis: muscle 0.4 g/kg/h plus liver up to 0.1 g/kg/h scaled by the liver-depletion input (set it low when carbohydrate was taken during the run, which spares liver glycogen; Jeukendrup 1999). The glycogen curve = absorbed glucose minus sinks with first call on it (obligate users 4.5 g/h; blood-pool refill ~4 g as a front-loaded exponential, τ 25 min; oxidation of 10% of absorbed glucose at the start rising to 25% with τ 90 min, because the post-exercise state stores glucose first: Krzentowski 1982), capped at that capacity; the surplus is oxidized or, later, stored as fat, and is shown as a tile. The saturating absorption rate is the rate at which that formula reaches capacity: (capacity + sinks) / (1 − oxidized share). Table cells turn green when absorption meets it.
Other foods and timing. Each row enters at its own time with its own lag and takes its share of the same total-energy cap, so its calories displace carbohydrate calories. Protein and fat slow the carbohydrate items' share through CCK/GLP-1 (up to −10% and −20%) without lowering the total; the brake follows protein and fat delivered to the small intestine in the trailing ~45 min, so it builds and fades smoothly rather than switching off when the stomach empties (Karamanlis 2007; Ma 2009). Solids compete by trituration state: a logistic in time since ingestion (inflection 40 min, plateau ~90 min) gives earlier food priority. Water leaves with a 13-min half-time and gives a dilution bonus that decays with it (Grimm 2025). Insoluble fiber lowers the solid-phase cap up to 10% (Benini 1995). Protein raises glycogen synthesis only when carbohydrate is below ~1.2 g/kg/h (Jentjens & Jeukendrup 2003); otherwise its value is muscle repair (20–40 g per feeding, repeated about every 3 h). A protein food 30–80 min into the meal brakes the carbohydrate (Ma 2009); the planner computes the 2-h carbohydrate cost by re-running the model without it. The magnitude by which a later solid delays an earlier one has not been measured in humans and is a model assumption.
Not validated. GI marked "est." has no direct measurement. No gastric-emptying study exists for koji amazake or corn grits; they are treated by food class. Modifier effect sizes come from small trials on standardized liquid meals, mostly in men; ginger combined with the post-run state has not been tested, and ginger's possible glucose-lowering effect is not modeled. The steady cap uses Hunt's mixed-meal mean rather than his volume and energy-density regression. Between healthy people, emptying varies about ±30%.
Sources. Hunt 1985; Krzentowski 1982; Berry 2003; Schirra 2000; Siegel 1988; Karamanlis 2007; Ma 2009; Gentilcore 2006; Giezenaar 2018; Collins 1988, 1991, 2002; Weiner/Meyer 1981; Huppertz & Chia 2021; Grimm 2025; Marciani 2012; Rehrer 1990; Ryan 1998; Keith 2018; Benini 1995; Margolis 2020; Areta 2013; Moore 2009; Hunt & Stubbs 1975; Brener 1983; Calbet & MacLean 1997; Vist & Maughan 1995; Kwiatek 2009; Datz 1987; Camilleri 2006; Wu 2008; Hu 2011; Lazzini 2016; Giacosa 2015; Zick 2008; Franke 2008; Jones & Horowitz 2000; Evans 2016; Yamaguchi 2026; Duchman 1997; Jeukendrup & McLaughlin 2011; Ijaz 2025; Cox 2010; Jentjens & Jeukendrup 2003; Burke, van Loon & Hawley 2017; Gonzalez 2016; Kurahashi/Oguro 2020; Atkinson 2021; Skrabanja 2001; Rokka 2014; Panlasigui 2010; AJCN 2022 oat β-glucan trial; Lin 1992.