Inference

ai-runner

MIT
Deploy

Batch AI pipeline runtime. Loads diffusion, speech, and language models onto the GPU and serves inference over REST.

By Livepeer · livepeer/ai-runner · contributors: victorges, rickstaa, yondonfu, leszko, j0sh

Pulls
384.4K
Stars
25
Last push
2026-03-24

Images

TagSizePushed
livepeer/ai-runner:latest10.9 GB2025-12-30
livepeer/ai-runner:audio-to-text14.4 GB2025-12-30
livepeer/ai-runner:text-to-speech14.4 GB2025-12-30
livepeer/ai-runner:segment-anything-211.5 GB2025-12-30
livepeer/ai-runner:llm9.1 GB2025-12-30

Endpoints

POST /text-to-imagePOST /image-to-imagePOST /image-to-videoPOST /image-to-textPOST /upscalePOST /audio-to-textPOST /text-to-speechPOST /segment-anything-2POST /llm
Run locally
Start the container and ping it to verify it’s serving.
docker run --gpus all -e PIPELINE=text-to-image -p 8000:8000 livepeer/ai-runner:latest
curl http://localhost:8000/health

Live from Docker Hub and GitHub, cached for an hour.