# My generation won’t run or failed

Why Studio refuses to start a card or a run, or a result fails: a model to choose, a key to add or check, the spending cap, the hourly limit.

![A Studio board with two cards that cannot run yet: Needs angle on the first, Choose 3 models on the second, and No model under their stages](https://docs.omnymous.com/media/getting-started/generate-your-first-content/board.5baaaaae.avif)

When Studio cannot start a run, a message in the corner says why, and a card that is not ready says what it still needs. Check in this order: the first causes are the most common.

## The card needs a model or an input

Every stage of a card runs on the model you choose for it: Studio never picks one for you. Until you do, the card reads **No model** under the stage, and **Choose a model**, or **Choose 3 models** for a card with three stages. A card can also wait for an input, as in **Needs angle**.

Open the card, choose a model for each stage under **Models**, and add what its inputs ask for. Until then the card cannot run. See [Generate your first content](https://docs.omnymous.com/getting-started/generate-your-first-content).

## A provider key is missing or was refused

Each model runs on your own key for its provider. Starting a run whose models have no usable key reads:

| What you see                                                                                                         | What it means                                                     | What to do                                                                                   |
| -------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------- | -------------------------------------------------------------------------------------------- |
| **A provider key is missing**: Connect a key for OpenAI in Settings → AI & Agents to run these models.               | No key is saved for that provider                                 | Add one: see [Add your AI keys](https://docs.omnymous.com/getting-started/add-your-ai-keys). |
| **A provider rejected its key**: OpenAI rejected the API key saved in Settings → AI & Agents, so this did not start. | The provider refused your saved key, and Omnymous switched it off | Check it, or replace it, below.                                                              |

Both messages have an **AI keys** button that opens the keys. A refused key reads **Rejected** on its card, and every run that needs it waits for it:

![The OpenAI card, Rejected: OpenAI refused it 1d ago · runs that need it are paused, with OpenAI’s words under Details, and Check again, Replace key and Remove](https://docs.omnymous.com/media/workspace/usage-and-ai-keys/rejected.64cee711.avif)

1. Press **Check again**. A provider having a bad minute refuses good keys too: **OpenAI accepts this key again** means the runs can start. **Couldn’t reach OpenAI to check the key** means it did not answer at all: try again in a moment.
2. If it still refuses, the key was deleted or revoked on the provider’s site. Create a new key there and press **Replace key**: see [Replace a key](https://docs.omnymous.com/workspace/usage-and-ai-keys#replace-a-key).

A key whose account has run out of credit is not switched off: its runs fail with the provider’s own words instead. Add credit on the provider’s site, then run the card again.

## The run is over its spending cap

Every run has a spending cap, $10 unless you set another, up to $250. A run whose estimate is over its cap does not start: **Over this run’s spending cap** says what it is estimated at, and the cap to set to run it. A run that reaches its cap while it runs stops there, and reads **Stopped at its spending cap**. Raise the cap and run the card again.

A single run takes at most 40 cards and 400 results. A bigger one reads **This run is too large**: run it in parts.

## You reached the hourly limit

A workspace starts at most 50 runs an hour, and makes at most 200 retries and AI edits an hour, counted apart: reworking what your runs made never spends your runs. Past either, a message says which limit is reached and when you can start again: **Hourly run limit reached** (**Your workspace has started its 50 runs for this hour. Try again at 2:05 PM.**), or **Hourly retry and edit limit reached**. While your runs are spent, the run confirmation holds **Run** and says when they start again. Runs already going are not affected, and typing your own edits is never limited by these.

## A result failed

A run can finish with some results failed: the failed ones read **This variation failed**, with the provider’s reason. Run the card again for new ones; what succeeded stays. If every result of a card fails with the same reason, the reason is usually its key: see above.

## Generation is paused

**Generation is paused for maintenance** means Omnymous has stopped everything that spends on your AI keys for a while: new runs, retries, AI edits, tracking a brand, refreshing a strategy map and rebuilding the brief. Runs and edits already going finish, typing your own edits still works, and nothing is charged for what was refused. Reconnaissance work already queued waits and runs once generation is back. Try again in a few minutes.

## Still stuck

Write to <support@omnymous.com> with:

- the board’s name and the card that failed;
- the exact words of the message, and when it appeared;
- the model and provider the card runs on.

Never send your API keys.

## Related

- [Run a board and control its cost](https://docs.omnymous.com/generation/studio/run-and-cost): Run one card, a card and what follows, a selection or the whole board; read what it should cost and set the most it may spend; then stop, finish or retry it.
- [Cards and models](https://docs.omnymous.com/generation/studio/cards-and-models): What a card on a Studio board shows, and its editor: how many it makes, the model each of its stages runs on, its inputs, its settings and its direction.
- [AI keys and usage](https://docs.omnymous.com/workspace/usage-and-ai-keys): Replace, check or remove the provider keys your workspace generates with, and read what they spent this month, by model and by type.
- [Generate your first content](https://docs.omnymous.com/getting-started/generate-your-first-content): Start a Studio board from a ready-made flow, give it an angle and models, run it, and find what it made in the Library.

---

Source: https://docs.omnymous.com/troubleshooting/generation-failed