A LaTeX editor with nothing left to install.

Turbotex ships its own TeX engine, language server, and Python kernel. Open a project, compile, and read the PDF beside the source. An agent drafts edits and plots figures when you ask.

Request early access Sign in

main.tex Compiled

\documentclass{article}
\usepackage{amsmath,graphicx}
\title{On the Convergence of Stochastic Descent}

\begin{document}
\maketitle
\section{Setup}
Let $f$ be $L$-smooth with step size $\eta$. Then
\begin{equation}
 \mathbb{E}\,f(x_{t+1}) \le f(x_t)
   - \tfrac{\eta}{2}\|\nabla f(x_t)\|^2 + \tfrac{L\eta^2\sigma^2}{2}
\end{equation}
\begin{figure}[t]
 \includegraphics{figures/loss.pdf}
 \caption{Training loss over 200 steps.}
\end{figure}
\end{document}

Everything a paper needs, already in the window.

The compiler is built in

Tectonic ships inside the app and fetches packages on first use. A fresh install compiles a real paper without TeX Live. Prefer pdflatex, xelatex, or lualatex? Point Turbotex at yours.
A preview that keeps its place

The PDF sits beside the source and stays on the same page across recompiles. SyncTeX links a line of source to its spot on the page and back.
\cite{kin
- kingma2015adam
- kingsbury2020
- kirkpatrick1983
A real language server

Texlab runs alongside the editor: completions for commands, labels, and citations, with diagnostics as you type. Errors from the compile log land on the line that caused them.
An agent that stages its edits

Ask for a rewrite, a new section, or a fix for the error in the log. Edits arrive as a diff you review before anything touches the file. It runs on your own ChatGPT account.
Figures from Python

Figure Maker runs matplotlib in a bundled Python kernel and writes the figure into the project. Keep the script when you want to regenerate it later.

A LaTeX editor with nothing left to install.

Everything a paper needs, already in the window.

The compiler is built in

A preview that keeps its place

A real language server

An agent that stages its edits

Figures from Python