# infr.sh

> Virtual inference for image-edit models. Realtime managed pricing (arena.ai leaderboard),
> GPU hosting for self-host, training cost tables, payment rails, and a routing proxy.
> **2% routed + standard fees apply** on inference calls. Catalog reads are free.
> Refreshed weekly. No auth required for read endpoints.

## Agent index

- [Manifest JSON](https://www.infr.sh/i/index.json): Map of every machine-readable endpoint
- [LLMs.txt](https://www.infr.sh/llms.txt): This file (mirror: https://www.infr.sh/i/llms.txt)

## Pricing — managed inference

- [Prices JSON](https://www.infr.sh/i/prices.json): Full catalog — 35 models, all provider cells, routed quote
- [Prices API](https://www.infr.sh/api/prices): Filterable JSON (?modality=edit&sort=price&routable=1)
- [Prices Markdown](https://www.infr.sh/i/prices.md): Human/agent-readable table
- [Single model](https://www.infr.sh/api/prices/{model-id}): One model + routed provider
- [Chain quote POST](https://www.infr.sh/api/prices/quote): `{ "steps": [{ "model_id": "..." }], "calls": 1000 }`

## Hosting — self-host GPUs

- [GPU JSON](https://www.infr.sh/i/gpu.json): Rental ($/hr spot/on-demand) + serverless ($/sec)
- [Hosting Markdown](https://www.infr.sh/i/hosting.md): Cheapest H100 spot summary

## Training

- [Training JSON](https://www.infr.sh/i/training.json): Trainable models + managed trainer prices ($/1k steps)
- [Training Markdown](https://www.infr.sh/i/training.md): Summary for LoRA / fine-tune planning

## Payments & credits

- [Onramps JSON](https://www.infr.sh/i/onramps.json): Peer ZKP2P, Stripe, Coinbase, MoonPay — 01.cash → Pute stack
- [Payments Markdown](https://www.infr.sh/i/payments.md): Back-office normalization model for agents
- [Payment rails JSON](https://www.infr.sh/i/rails.json): Tempo, x402, Stripe, onramps — fees and limits
- [Deposit UI](https://www.infr.sh/deposit): Human onramp flow (Peer extension + redirects)
- [Pute credits](https://www.infr.sh/pute): 01.cash primitive + prepaid compute balance
- [Credit sharing JSON](https://www.infr.sh/i/sharing.json): P2P, gifting, team pools across providers

## Routing — run inference through infr

- [Routing guide](https://www.infr.sh/i/routing.md): POST https://www.infr.sh/api/v/{model-id} — Privy credits, $0.20 trial, MPP/402
- [Planner POST](https://www.infr.sh/api/orchestrate): Describe a goal → primary chain + 2 alternatives

## SDK & CLI

- npm package: `infr`
- CLI: `npx infr prices`, `npx infr price {id}`, `npx infr quote {ids}`, `npx infr plan "goal"`

## Human UI

- [Price table](https://www.infr.sh/): Sortable/filterable browse
- [Build endpoint](https://www.infr.sh/build): Plan chains + ship
- [Docs](https://www.infr.sh/docs): Full API reference

Updated: 2026-05-28 · 2% routed + standard fees apply