What does your metabolic model actually predict?

Paste a small metabolic model. Your browser runs flux balance analysis exactly as cobrapy does - the optimum, pFBA fluxes, flux variability, blocked reactions, gene and reaction knockouts, mass balance - free, nothing uploaded. A paid run then interprets the result or writes the cobrapy script that reproduces it.

Each example has a saved model run, so you can see the whole page for free.

Analysis

Model (one reaction per line - ID: equation [lower, upper] gpr: rule - plus objective: ID; or paste a cobrapy JSON model)

Drop a .json (cobrapy) or .txt model, or

Arrows follow cobrapy: --> is [0, 1000], <=> is [-1000, 1000], <-- is [-1000, 0] unless you write bounds. A reaction with one metabolite is a boundary (exchange) reaction. Optional metabolite: id | formula | charge | name lines enable the mass-balance check. SBML is not read - export it with cobra.io.save_json_model.

Settings

FVA at cobrapy's fraction_of_optimum; the other analyses always use the optimum itself.

Notes (optional: the organism, the medium, what the objective stands for, what you want to conclude)

Load set .json

Flux balance analysis (free, in your browser, no AI)

Metrics

Id	Metric	Value	Basis

Exchanges in the pFBA state

Reaction	Metabolite	Flux	Direction	Bound

Largest internal fluxes

Reaction	Equation	pFBA flux	FVA range

Flux variability

Reaction	Min	Max	Kind

Knockouts

Deleted	Kind	Objective after	Status

Structure: blocked reactions, dead ends, mass balance

Flags

Free: no account, no run.

Anything to ask (optional)

Paste a model to price the run.

Raw model reply

The reply did not parse as the structured result, so it is shown as it arrived.

Your recent runs

What this does, and what it does not

Flux balance analysis finds the flux through every reaction that maximises (or minimises) an objective - usually a biomass reaction - while every metabolite is at steady state (S v = 0) and every flux stays within its bounds. The page solves these linear programs with its own simplex method and follows cobrapy 0.32.1 for every definition: slim_optimize, pfba, flux_variability_analysis, find_blocked_reactions, find_essential_reactions and find_essential_genes (below 1% of the optimum, or infeasible) and check_mass_balance. It was checked against cobrapy 0.32.1 with GLPK on the E. coli core model and on hundreds of random models and bound changes; the only differences found were deletions that land exactly on the 1% threshold, which solver rounding decides either way and which the page flags.

An optimum is a property of the model and its bounds, not a measurement: pFBA picks one of possibly many optimal flux states, FBA knows nothing of kinetics or regulation, and a result that moves when the default bound of 1000 moves is an artefact of that cap. The paid run reads only what the browser computed and your notes; it is told never to compute a new number, and the page checks every number it writes. Derived from the agent skill @k-dense-ai/cobrapy (k-dense-ai/scientific-agent-skills, K-Dense Inc.). The E. coli core model is the one cobrapy bundles (Orth, Fleming and Palsson; see the notice); the other example is a made-up teaching network.