Geometry diagram maker: questions and answers
Last updated 28 October 2026
makegeodiagrams.com draws the diagram of a Euclidean geometry problem from the problem's own statement, checks the figure against every condition the statement gives, and lets you restyle it, edit it by hand or in words, move its points, and download it as SVG, PDF, PNG or Asymptote code. The questions below are the ones people ask before and after their first figure. The homepage shows the chat at work and the gallery shows finished figures you can drag.
What it is
What does makegeodiagrams.com do?
It draws a geometry diagram from a problem statement. You paste the statement of a Euclidean (plane) geometry problem as written, in plain text or LaTeX. The statement is translated into a construction in which every point, line and circle is defined by how it arises from the others; the construction is solved numerically for a figure in which every stated condition holds; the conclusion is checked on that figure; and the figure is drawn by Asymptote, the way figures are drawn for mathematics books and competition booklets. You get a diagram you can restyle, edit, move live, and download as SVG, PDF, PNG or Asymptote source.
How do I draw a geometry diagram from a problem statement?
Open the app, which asks you to sign in or make a free account, paste the complete statement exactly as written into the chat, and send it. After a few seconds the reply shows the figure and lists what was checked: each condition of the problem and each claim, with whether it holds in the drawn figure. From there you can change the look with Effects (one click for grayscale, colour or focus), ask for changes in words, open Edit design for hand control, drag the points with Move, and download the figure in the format you need. Every version stays in the conversation.
Who is it for?
Anyone who needs a correct figure of a geometry problem rather than a sketch: writing up a solution, making a worksheet, a handout or an exam, illustrating a book, a paper or a set of olympiad notes, posting a problem on a forum or a blog, preparing a lesson, or understanding a problem by moving its points. Students, teachers, authors and problem setters use it the same way.
Is it an AI geometry diagram generator?
Partly, and the part that matters is not AI. Reading the statement and writing the construction is done by AI models; everything after that, solving the construction, checking the conditions and the conclusion, placing labels and marks, and drawing, is deterministic mathematics. That is why a figure here is checked rather than guessed: an AI-drawn picture can look right and be wrong, while here a figure that fails a stated condition is never shown as if it held.
Is it a geometry solver, a proof assistant or a construction tool like GeoGebra?
None of the three. It does not solve or prove the problem: the conclusion is tested numerically on the figure that was drawn, and a claim that holds in one figure is evidence, not a proof. You do not construct the figure yourself either, as you would in GeoGebra or Desmos Geometry: the construction is read from the statement. Afterwards the figure behaves like a dynamic-geometry figure, drag any point and the whole construction follows, and the design editor has construction tools (points, lines, circles, intersections, triangle centres, conics, marks, text) for anything you want to add.
Which geometry problems can it draw?
Plane geometry: points, lines, segments and rays, circles and arcs, polygons, conics (ellipses, parabolas and hyperbolas), loci, and the classical named objects such as incircles and excircles, the nine-point circle, the Euler line, symmedians, mixtilinear circles and the Miquel point. That covers olympiad problems, textbook problems and contest problems with given lengths and angles. It does not draw three-dimensional figures, and a problem has to describe a figure: a counting problem about lattice points has no diagram to draw.
Can it draw olympiad problems, such as IMO or shortlist problems?
Yes; that is what it was built for. The gallery shows finished figures of problems from the IMO, its shortlists and the Sharygin olympiad, each one live: IMO 2008 Problem 6 (a circle tangent to four lines of a quadrilateral) and IMO 2019 Problem 6 (the incircle and a perpendicular through its touch point) are typical. Each figure there was drawn from the statement in the same chat you use; the olympiad geometry page says what such a figure has to get right.
What happens if the statement is wrong, incomplete or impossible?
The figure follows the words. A missing or mistyped condition draws a different figure, or none. Conditions that cannot all hold at once are reported as having no figure; the site never shows a figure of a different problem in place of yours. Text around the problem (a source line, a problem number, chatter) is set aside and not drawn, and the reply says so. Check the statement before you paste it: a wrong statement still uses credits, though on the paid plans a new figure that cannot be drawn at all is free within limits, which the plans page gives.
Putting a problem in
Can I paste the problem in LaTeX or in plain text?
Both, as written, copied from a PDF, a web page or a forum post. Notation is read in any spelling: $\Gamma_1$, Γ₁ and Gamma1 are the same circle, \angle ABC and ∠ABC the same angle, and the chat shows the statement typeset in standard LaTeX. Named objects are labelled in the figure exactly as the statement names them.
Can I give the problem as a picture or a PDF?
Not as a picture: the site reads text. Copy the statement out of the PDF, or type it from the image; a phone's text-recognition camera or any OCR tool gives you text to paste. Formulas may come through as plain text or as LaTeX, and both are read.
Can I get a figure for a solution as well as for the problem?
Yes. Attach the solution to its problem from the chat's message box and the solution gets a figure of its own, with the points, lines and circles the solution introduces drawn under the solution's own names: the figure you would print next to a solution in a book, with the conclusion of the problem marked and the auxiliary objects of the argument present.
Does it draw locus problems?
Yes. A locus problem comes back with the locus traced on the printed figure, the reply says along which object of the figure the traced point runs, and in the live figure the locus is drawn again as you move the given points. Effects has a Show locus switch for printing the figure with or without the trace.
Can I change the figure by asking in plain English?
Yes. Write what you want in the chat: draw the altitude from A and call its foot D, mark the two equal angles, make the circles red, hide the label of P, crop the circumcircle to the arc the figure uses, rotate the figure so that BC is horizontal. A change to the construction is checked again like a new figure; a change of style is applied at once. Each change is a new version, and the earlier versions stay in the conversation. Most edits take a couple of seconds and one or two credits.
What you get
What file formats can I download?
SVG and PDF (vector, sharp at any size, for print and LaTeX), PNG at 300 dpi and at 600 dpi (for documents, slides and posts), the Asymptote source that compiles to the same figure, and the figure as a live web page in which readers drag the points. All of them are Asymptote's own output from one source, so they match to the pixel. Which of these a plan includes is on the plans page: the Free plan downloads PNG; the paid plans add the vector files and the source; live web pages come with Pro.
How do I put the diagram in a LaTeX document?
Download the PDF (or the SVG) and include it with \includegraphics, as you would any figure. Or download the Asymptote source: with the asymptote package the code goes into an asy environment inside the document, and on its own it compiles with asy -f pdf to the same figure. The labels are typeset by TeX in Computer Modern, so the figure matches the text of a LaTeX document. The Asymptote page has the commands.
Can I get the Asymptote code of the figure, and is it usable?
Yes, on the Personal and Pro plans. It is the figure's own source, the one Asymptote compiled, cleaned for reading: the points are named pair variables, circles are paths, each label is written relative to its point, only the pens the drawing uses are defined, and a comment says what each object is. It compiles with Asymptote on its own or inside a LaTeX document and gives the same figure, and you can keep editing it by hand.
Can I get TikZ code?
No. The site produces Asymptote code, SVG, PDF and PNG. A document that draws its other figures with TikZ can include the PDF or the SVG as a graphic, and the Asymptote source can be compiled beside the document; there is no conversion to TikZ.
How are labels, ticks and angle marks placed?
Labels are placed on the finished drawing, clear of the ink, where each one reads as its own point's name. Equal segments get ticks, equal angles arcs and right angles their squares only where the statement states them or you add them. What is to be proved is drawn apart, dashed and in its own colour, so that a conclusion never reads as a given. The printed size is chosen for each figure, from about 7 to 16 cm, the way a contest booklet sizes its figures, and you can set it yourself.
How do I change colours, line widths, labels and marks?
With Effects for one-click looks: Grayscale for black-and-white print, Colorful, Focus (the stated objects heavier, helper lines dotted), No fill, Crop big circles, Show locus. With Edit design for hand control: line widths, colours, dash styles, arrowheads, point markers, label positions and sizes, marks and their counts, fills, measurements, LaTeX text anywhere, new constructions, and your own styles saved as named style sets. And by asking in the chat. Any version can be downloaded.
Can I embed an interactive geometry figure in a web page or an online book?
Yes, with Pro. The Live download gives a zip with a small web page, the figure's data and the player; put the folder on your site and embed the figure with one line of HTML or an iframe. Readers drag the points, pin objects, play the animation you recorded and double-click to return to the start; whatever you pinned stays pinned; nothing is sent to any server. The gallery's figures are the same component.
Correctness and live figures
How do I know the diagram is correct?
Every stated condition is a constraint the figure satisfies to numerical precision, and the reply lists what was checked: each claim of the problem, and whether it holds in the drawn figure. A claim that fails is reported and never drawn as if it held. A figure in which points coincide or a triangle collapses is refused and another is searched for, and the figure shown is a generic one, never the symmetric special case the statement did not ask for.
Can I drag the points, like in dynamic geometry software?
Yes. Move turns the figure into the construction solved in your browser: drag any point and every other point, line and circle follows so that every condition of the problem still holds; every position you see is a true figure of the problem, not an in-between picture. You can pin points, lines or circles so that they stay, let a point run along its circle or line, record an animation, and go back to the start with a double-click. Save the moved figure and it becomes a new version.
Can I get a different layout of the same figure?
Yes. Say draw another figure and you get a genuinely different drawing of the same construction, never a near-copy, with the layout that still holds the conclusion preferred. Move lets you drag the points to the layout you want, Make horizontal turns the figure so that a chosen line is level, and Edit design rotates it by any angle. Each is saved as a version of its own.
How long does a figure take?
Most figures take a few seconds. The hardest problems can take a few minutes, and the chat shows the time and the credits used while it works, with a Stop button that keeps the best figure found so far.
Does it work on a phone or a tablet?
Yes. On a phone the chat and the diagram are two screens, the diagram's tools sit along the bottom within reach of a thumb, and points are dragged with a finger in Move and in the gallery. Every page of the site is laid out for a phone as well as a desktop.
Plans, credits and licences
Is it free?
There is a Free plan: a monthly allowance of credits, PNG downloads at 300 and 600 dpi, and the chat, Move, Effects and the design editor. The paid plans give more credits each month, SVG, PDF and the Asymptote source, and, on Pro, live web pages and commercial use. The plans page has the numbers.
What is a credit, and what uses credits?
Credits measure the work behind a figure. A new figure and a chat edit use credits: most new figures a few, an edit one or two, a hard problem that needs several attempts more, and no single action can use more than a limit you can lower on your account page. Editing the design, moving points, Effects, size changes, downloads and redraws use none. If the credits left may not be enough for a figure, the chat asks before it starts.
Can I use the figures in a book, a paper, a course or a post?
Figures made on the Free and Personal plans may be used for study, teaching and posts you earn nothing from; figures made on Pro may be used in anything, including books, papers, paid courses and commercial material. The copyright page is the full statement. The problems and solutions you send stay yours.
Privacy
What happens to the problems I send?
Your conversations are never opened by us (unless you send one as feedback, ask us to look at it, or the law requires it), never published and never used to test or improve the service. Deleting a conversation deletes it from the server at once, and the server logs never contain what you write. One limit: to be read, a statement is sent to third-party AI service providers, and one of them may use what it receives to improve its own models, so a problem that must stay secret, such as a competition proposal, should not be sent. The privacy page says exactly what leaves the server.
Technical
What draws the figure?
The final figure is drawn by Asymptote, the vector graphics language used for mathematical figures in LaTeX documents and on mathematics forums, with labels typeset by TeX; the SVG, PDF and PNG files are Asymptote's own output from one source. The construction language, the numerical solver that finds the figure, the label placement and the solver that runs the live figures in the browser were written for this site.
How is this different from GeoGebra?
In GeoGebra you build the figure yourself, step by step, with its tools; here the figure is built from the problem's text, and the construction tools are there for what you want to add afterwards. GeoGebra is a general dynamic-geometry environment; this site does one thing: the diagram of a stated problem, checked against the statement and printed as it would be in a book, with the Asymptote source to go with it.
How is this different from asking an AI chatbot to draw the figure?
A chatbot's picture is its impression of the text: it can be convincing and still break a condition of the problem, and nothing checks it. Here the statement becomes a construction, the construction is solved so that every condition holds, the conclusion is tested on the drawn figure, and the reply says what holds and what does not. What you get is a compiled Asymptote figure with its source, and a live figure that keeps every condition while you move it.
More
The homepage shows a figure being made; the gallery has finished figures to drag; the Asymptote page shows the code and how to use it in LaTeX; the plans page has the plans and their own questions; the copyright and privacy pages are the full statements. Anything else: contact@makegeodiagrams.com.