The models we watch, side by side

Welcome to
AiSussed Orbit.

What each one is for, what it costs, what changed — and, when we are not sure, we say so instead of guessing. Pictures, video, music and a voice are in here now, next to the chat models. Midjourney, ElevenLabs, Suno and Runway are not, because none of them publishes a list anybody could check.

A human typed this in on 24 September 2026

ORBIT / THE NAVIGATOR

I brought the map. Everything in here has a date on it and a link to where it came from. When it says “we have not checked this”, that is me being honest, not me being lazy.

  • Anthropicwe checked under an hour ago · 3 changes we are still checking
  • Googlewe checked under an hour ago · 1 change we are still checking
  • OpenAIwe checked under an hour ago
AI tools in here
6
Companies that make them
3
We last looked
within the hour
Changes we are still checking
4
Provider
Motion is reduced.
Takes or makes
View

6 models shown across 3 providers.

6 toolsone per orbit, grouped by who makes them · hover or tap to stop the spinning and pick one

Anthropic

2 models

Google

2 models

OpenAI

2 models

Start from the job

What do you actually want it to do?

Pick a job and we will show you the tools that claim to do it. We only count tools where we have checked this ourselves, so the numbers are small and honest rather than big and hopeful.

What changed

The most recent records.

Every record says when it was written and where it came from. Nothing here was inferred.

  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026
  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026
  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026
  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026
  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026
  • Source listing captured; live monitoring pendingTyped in by a human · written 26 Sept 2026

Reading a record

What the labels mean.

Every card tells you where its details came from, so you know how much weight to put on them.

Verified from official list

The provider’s own API listed this model, an editor checked the change, and only then did it become public. The date on the card is when we read the list.

Editorially corrected

A person wrote or fixed this after reading the provider’s documentation. The reason is recorded and shows in the model’s history.

Curated preview

Written by hand before live monitoring covered this model. Useful, but the least evidenced of the three. Check the linked source.

Not yet verified

We would rather leave a gap than guess. A blank context window or availability means nobody has confirmed it, not that the model lacks the feature.

How this stays current

What we check, and what we don’t.

Official model lists, hourly

We read each provider’s own API and compare it with what we last saw. A scan that suddenly drops a large part of a catalogue is treated as a broken response, not a mass retirement.

Running for Anthropic, Google, OpenAI.
Release notes and pricing pages

We watch the official pages for change and record what moved. We do not read a price off a page and publish it; a person records pricing from the source, with the date.

Nothing publishes itself

Every change waits in an editorial queue. A model is only marked retired after a person confirms it, and a link only appears after it has been fetched and answered.

What we cannot tell you

Whether a model is right for your particular job, how it will behave next month, or anything a provider has not published. Where we do not know, the card says so.

Orbit is only the beginning

Now: ask every model
the same question.

The Arena is open. Your question goes to all of them at once, the names stay hidden until you have said which answer actually helped, and most people find they cannot tell them apart.Try the Arena ↗