Livepeer

Decentralized video infrastructure and AI video on Ethereum + Arbitrum.

Livepeer is the open, permissionless protocol for video transcoding, streaming, and AI video inference, coordinated on Ethereum mainnet and Arbitrum One via the Livepeer Token (LPT). The company commercializes the network through Livepeer Studio — a managed REST API for live streaming, on-demand video, multistream, rooms, access control, recording, metrics, and AI generate — and the open Livepeer AI Network, where any GPU operator can serve text-to-image, image-to-video, live video-to-video, LLM, audio-to-text, text-to-speech, upscale, and segmentation pipelines for fees settled in ETH via probabilistic micropayments.

This repository is the API Evangelist catalog entry for Livepeer.

APIs

This catalog covers three distinct API surfaces:

API	Base URL	Auth	Operations
Livepeer Studio	https://livepeer.studio/api	Bearer (API key)	64
Livepeer AI Network (community + Studio gateway)	https://dream-gateway.livepeer.cloud / https://livepeer.studio/api/beta/generate	Bearer (API key)	13
Livepeer Node CLI HTTP (go-livepeer local API)	http://127.0.0.1:7935	Loopback	12

Total: 89 operations across 4 OpenAPI specifications.

Livepeer Studio surface (by tag)

• stream — 11 ops (create, retrieve, update, delete, terminate, start-pull, multistream targets, clip)
• asset — 6 ops (list, request-upload, upload-by-URL, retrieve, patch, delete)
• webhook — 8 ops (CRUD + logs + resend)
• multistream — 5 ops (target CRUD)
• session — 4 ops (list, retrieve, recorded sessions, clips)
• room — 9 ops (WebRTC rooms + users + egress)
• transcode — 1 op (file transcode)
• playback — 1 op (retrieve playback info)
• accessControl — 5 ops (signing key CRUD)
• task — 2 ops (list, retrieve)
• metrics — 5 ops (realtime + historical viewership, usage)
• generate — 7 ops (text-to-image, image-to-image, image-to-video, upscale, audio-to-text, segment-anything-2, llm)

Livepeer AI Network surface

10 generation pipelines:

• POST /text-to-image
• POST /image-to-image
• POST /image-to-video
• POST /upscale
• POST /audio-to-text
• POST /segment-anything-2
• POST /llm (OpenAI chat-completion compatible)
• POST /image-to-text
• POST /live-video-to-video
• POST /text-to-speech

Plus runner introspection: /health, /hardware/info, /hardware/stats.

Livepeer Node CLI HTTP

12 endpoints exposed by go-livepeer for operators: status, protocol parameters, registered orchestrators, bond / unbond / rebond, orchestrator activation, broadcaster config, max price per capability, reward, ETH/LPT transfers, message signing.

Repository contents

Folder	What lives there
apis.yml	APIs.json 0.19 entry indexing every API and artifact
openapi/	OpenAPI 3.1 specs for Studio, AI Worker, AI Gateway, CLI
rules/	Spectral rulesets enforcing Studio + AI conventions
capabilities/	Naftiko capability YAMLs (one per business surface)
vocabulary/	Domain vocabulary covering protocol, Studio, AI, tokens
json-ld/	JSON-LD context for Livepeer resources
json-schema/	Stream, Asset, Task, Webhook, AI pipeline schemas
json-structure/	Operational structures (platform, stream)
examples/	Example request/response payloads
plans/	API Commons 0.1 Plans (Sandbox / Growth / Enterprise)
rate-limits/	API Commons 0.1 Rate Limits
finops/	FOCUS-aligned FinOps mapping
review.yml	API Evangelist apis.yml lint review

Why Livepeer matters

• Tier-1 decentralized infrastructure provider: The largest non-financial public good built on Ethereum that delivers measurable user-facing utility (video).
• First production AI compute market: Livepeer's AI Subnet is one of the only crypto networks where AI inference workloads, not speculation, drive on-chain fee revenue (~60–72% of fees per recent quarterly metrics).
• Two posture options for the same workload: Builders can start with the managed Studio API (familiar SaaS, fiat billing) and graduate to the open Gateway when they want to choose Orchestrators, pay in ETH, and operate trust-minimized.
• Open governance: LPT holders steer treasury spend, protocol parameters, and upgrades via LIPs.

Naftiko capability surface

Every API surface is decomposed into shared per-domain Naftiko capabilities, each exposing:

• A REST adapter (one resource per consumed OpenAPI path)
• An MCP adapter (one tool per consumed operation)

These capabilities are designed to be composed into governed AI workflows (e.g., "ingest creator livestream → record session → transcode VOD → generate text-to-image thumbnail → push to multistream"). Authentication is bound to LIVEPEER_API_KEY via environment.

Maintainer

Kin Lane — kin@apievangelist.com

---

Generated and curated by API Evangelist. Last refreshed 2026-05-25.