Bolly for Codex

Generate on Bolly,
from Codex.

Add Bolly as an MCP server in the Codex CLI and generate images, video and audio with every model in the catalog, straight from the terminal.

Get an API key

Three steps, no provider keys.

Under a minute from copy to first run in Codex.

1

Create an API key

Open Settings → API keys on Bolly and create a key. Copy it — it's shown only once.
2

Add it to ~/.codex/config.toml

[mcp_servers.bolly]
url = "https://www.bolly.ai/api/mcp"
http_headers = { "Authorization" = "Bearer YOUR_BOLLY_API_KEY" }
3

Ask for something

Restart Codex and ask: "generate a hero image for this landing page with Nano Banana 2". Codex calls run_model and hands you the URL.

What Codex can do with it.

Find

Picks the right model

Describe the task and the assistant searches the catalog: remove a background, turn an image into video, upscale. Same models as the app.

Run

Runs it on your plan

Runs are charged to your Bolly plan, at the same price as in the app. One API key, no provider accounts to manage.

See

Hands you the result

Every run returns direct URLs for the image, video, audio or 3D file, plus a link to open it in your Library.

Chain

Picks up where you left off

Results stay in your Library, so "make a video from that image" works across sessions.

Seedance 2.5, Nano Banana 2, Minimax H3, GPT Image 2 and every other model on Bolly. See the full list

Frequently asked.

Which Codex versions are supported?+

Any Codex CLI release with remote MCP support (mcp_servers with a url and http_headers).

What does it cost?+

Runs are billed to your Bolly plan at the same credit prices as in the app.

Where do the results go?+

Every run lands in your Library, and Codex gets direct URLs to the generated files.

How do I disconnect?+

Remove the block from config.toml, or revoke the key on the API keys page.