Our models
Theo orchestrates frontier engines from several labs and shows which one answered each step. A few engines are different: OpenCharts builds them or runs them on its own hardware, and each has a public page that says what it does, what it guarantees, what was measured and when, and where it is available. No sign-in needed: open opencharts.com/models.
What it is
The hub lists 4 engines. Each one is a model OpenCharts runs on its own hardware, or through its own lab, and stands behind by name. The pages are deliberately plain: a capability list in plain words, what happens to your request step by step, every number with its date and source, the safety and retention controls, and the facts of the deployment: where it runs, how it is versioned, and what is kept.
Every name and number on the hub is data from one registry that is held to the product's own registries by a contract test, so what the hub says about an engine is what the product calls it.
The roster
Theo Moonlight 1.1
Live in TheoImage engine. High fidelity image renders on our own hardware.
Lives in Theo Stealth. Read its page
Theo Real World 1.1
Rolling outCapture engine. Photos of a real thing become a 3D capture of the thing in its space.
Lives in Theo Chat, Real World mode. Read its page
TheoArca Magnus 5.1
Rolling outText engine. Theo's own deep-reasoning engine, governed and audited.
Lives in Theo Chat and Code Canvas. Read its page
TheoArca Velox 5.1
Rolling outText engine. Theo's own low-latency engine, governed and audited.
Lives in Theo Chat and Code Canvas. Read its page
Reading a model page
What it does and how it works
The capability list is written for a reader, not a benchmark table, and the pipeline is told step by step: what happens to your prompt or your photos before, during and after the model runs.
Measured, with dates
Every number is our own measurement on our own hardware: a render time, a capture duration, a human review round. Each says when it was measured and where it came from. None is a benchmark score.
Safety and retention
The controls in front of and behind the model, stated as mechanisms: what is screened, what fails closed, what is kept, for how long, and why. Where a model keeps anything, the page says so in as many words.
Where it runs and where to find it
The deployment facts as rows: the hardware it runs on, how it is versioned, what happens to what you send it, and what is kept. Where to find it lists the Theo surface the model lives in and every other channel, marked live, in progress or planned, linking only to what exists.
Models and the benchmarks page
The text engines on the hub also appear on the benchmarks page as first-party rows. Their scores there are OpenCharts' own runs of the public tests and stay labelled internal, independent verification TBD, until a third party publishes a score; they never lead a spotlight. The image and capture engines are not ranked there: the OpenCharts Index ranks text models on third-party published tests, and there is no equivalent for an image or a capture engine yet.
Using a model inside Theo
Each page's Where to find it section names the Theo surface the model lives in and how to pick it there: an engine in the Stealth Engine menu, a mode in the Create row, the composer's Theo Arca pick, or a Code Canvas tier. Some are rolling out and appear only where they are enabled; the page says so rather than promising a control that is not there.
For agents
The whole hub is published from the same registry the pages render:
- /models/llms.txt as plain text, in the llms.txt style AI assistants read.
- /models/models.json as JSON with every model, its measured facts (dated, sourced), safety controls, availability and distribution status.
Quote with a link to the model's page.
FAQ
Are these the engines Theo usually answers with?
No. Theo is an orchestrator and shows which lab engine handled each step. The models on the hub are the ones we build or run ourselves, and each page says exactly where it appears.
Does a page say how an engine is built?
No. A page describes what an engine does, what it guarantees and what we measured, not how it is assembled. How our models are built is our own work and stays with us; what you can hold us to is written on every page: the safety controls, the retention, the versioning and the dated numbers.
Can I call these models through an API?
Not yet. Every page carries a Where to find it section that marks what is live, in progress or planned and links only to what exists. When a channel goes live, it is linked there.
Do I need an account?
No. Every models page and both machine-readable files are free and public. Using a model inside Theo follows that surface's own plan rules.
Inside a chat, Theo always shows which engine answered each step. The Theo Engines article explains how that trace reads.Related Articles
Theo Engines
Every Theo-branded engine powering the product — generated directly from the unified model catalog.
Theo Stealth (Private Mode)
Zero-retention private AI mode — ephemeral conversations, uncensored image & video generation, no history or memory.
Real World
Turn photos or a short walk-around video of a real person, object, or space into a faithful 3D capture of the thing in its surroundings — a quick capture in about a minute and a full one a few minutes later, on one card. Spin it, walk through it full screen, or view it in AR. Nothing is invented.
Benchmarks
The free, public ranking of every notable AI model by the OpenCharts Index — a category-balanced average of third-party benchmark scores measured against the best published result on each test — with letter grades, rank bands, category leaders, a hard-set score, a page per model and per benchmark, provenance labels on every number, and llms.txt / JSON files for agents. No sign-in needed.