Cerebras Backend

Cerebras provides fast inference on custom hardware. Uses the same multi-frame batch strategy as OpenAI.

Setup

1. Get an API key from cerebras.ai
2. Run eyeroll init and select Cerebras, or set it directly:

export CEREBRAS_API_KEY=your-key-here
export EYEROLL_BACKEND=cerebras

Capabilities

Feature	Supported
Direct video upload	No
Multi-frame batch	Yes (all frames in one API call)
Audio transcription	No
Text generation	Yes
Preflight health check	Yes

Model

Override with the --model flag:

eyeroll watch video.mp4 --backend cerebras

How it works

Cerebras uses the OpenAI-compatible API format. eyeroll sends all extracted frames as base64-encoded images in a single API call, with timestamps per frame. No audio transcription is available.

Cost

Competitive pricing with fast inference. Check cerebras.ai for current rates.

Install

pip install eyeroll[openai]

Uses the openai SDK with a custom base URL.