Put an Inliner URL in HTML, email, docs, or an agent workflow.
The first request generates the image into your project; every request after serves it from Cloudflare's edge.
<img src="β¦"> is the whole integration.
.webpUpgraded in place. FLUX 3 handles default generation across 14 native aspect ratios; Nano Banana 2.1 locks subject identity across reference images and edits.
Model routing guide βFlare renders crisp packaging, UI layouts, and lettering up to 2048Γ2048. Sunburst handles local edits with optional native rectangular context masks.
Mount the Studio editor inside your CMS or SaaS app via iframe or InlinerEditor.embedIframe() and receive published lossless WebP URLs over postMessage.
01 How it works
Write the image path you need in your markup. The first request renders the asset into your project namespace; every request after comes straight from Cloudflare's CDN.
Include your project slug, a hyphenated description, and pixel dimensions. Write it by hand, from a build script, or let your coding agent insert it.
Inliner routes the prompt to the best-fit model (or the one you specify) and applies your project's saved brand instructions and reference images.
Once generated, that URL is pinned to the finished image and cached globally. Append .webp anytime to serve a compressed WebP copy on the fly.
A generated URL always returns the same image, so it is safe to commit into production templates and transactional emails.
Use the exact same URL in your web app, CMS, newsletter, and README without exporting separate files per destination.
Assets stay grouped by project namespace with automatic vision tags, alt text, and search.
02 Who ships with it
From local dev servers and coding agents to client email templates and CMS image fields.
Write the image you need directly into your markup. It resolves to a real, hosted asset on first load β with SDKs for JS, Python, Go, and PHP when you outgrow the URL.
Claude Code, Cursor, Codex CLI, Copilot, and any MCP-compatible tool can generate, edit, and manage hosted images without leaving the coding session.
Email-safe hosted visuals for every client and every send. Per-client project namespaces, shared credits, and URLs that never rot in an old template.
Paste an Inliner URL into Webflow, Framer, Carrd, or any CMS image field. Describe what you want; the asset appears and stays hosted.
03 Build workflow
Start with a plain <img> tag. Add the CLI, MCP server, embeddable editor, or typed SDKs when your workflow needs them.
.webpIf a surface can load a URL, it can render an asset. Append .webp to any existing PNG or JPG URL to serve WebP on the fly at zero credit cost.
Generate and edit from the shell. Prints URLs in interactive terminals and raw image bytes when piped. CLI docs β
Expose generation, editing, project namespaces, and credit checks to Claude Code, Cursor, and Codex. Setup guide β
Mount the Studio layer and AI editor inside your app and receive saved WebP URLs over postMessage. Editor overview β
Official JavaScript, Python, Go, and PHP clients for server-side generation, search, and tag management. API reference β
<!-- The URL is the whole integration. --> <img src="https://img.inliner.ai/acme/spring-campaign-hero_1200x600.png" alt="Spring campaign hero" /> <!-- Same mechanic in email templates --> <td background="https://img.inliner.ai/acme/newsletter-header_600x200.png"> <!-- Edit by appending an instruction --> <img src="https://img.inliner.ai/acme/spring-campaign-hero_1200x600.png/make-it-night-time.png" />
$ npm install -g @inliner/cli $ inliner-ai "product-launch-banner_1200x400.png" -p acme https://img.inliner.ai/acme/product-launch-banner_1200x400.png $ inliner-ai edit hero.png "remove the background" # pipes work like any unix tool $ inliner-ai "og-card_1200x630.png" -p blog | pbcopy
$ claude mcp add inliner -- npx -y @inliner/mcp-server # then, inside your coding agent: > "Add a hero image to my landing page" > "Generate three variants of the pricing illustration" > "Edit the team photo to remove the background" # the agent gets real hosted URLs back β # no downloads, no asset handoff, no manual upload
<div id="inliner-editor"></div> <script src="https://app.inliner.ai/inliner-editor/inliner-editor.iife.js"></script> <script> InlinerEditor.embedIframe('#inliner-editor', { sourceUrl: 'https://img.inliner.ai/acme/hero_1200x600.png', project: 'acme', editModel: 'quality', // GPT Image 2.5 Sunburst useContextMask: true, // native rectangular context mask onPublish(p) { heroImg.src = p.url; } // lossless .webp CDN URL }); </script>
import { InlinerClient } from '@inliner/sdk'; const client = new InlinerClient({ apiKey: 'inl_β¦' }); // generate a hosted asset const img = await client.generateImage({ project: 'web', prompt: 'empty state illustration, warm minimal', }); // list, search, and manage the library const { items } = await client.listImages({ project: 'web', tags: ['illustration'], });
04 Asset library
Every generated or uploaded image lands in a project namespace with vision-generated tags, editable alt text, and role-based team access.
Invite teammates with owner or member roles, draw from a shared credit pool, and issue inl_ keys for CI scripts, apps, and MCP agents.
Team accounts βVision AI writes tags, titles, captions, and alt text on creation. Filter across multi-tag queries or rename slugs from saved image context.
Asset management βIsolate clients or environments under their own URL prefix. Attach default prompt instructions and up to five reference images per project.
Namespace workflows β05 Edit & resize
Every hosted image can branch into a new variant. Append a plain-English instruction or target pixel size to render a derivative at its own permanent URL.
Describe the change. Recolor an object, swap a background, remove an element, or restyle the whole scene.
Choose the dimensions. Request the exact aspect ratio and pixel size your page, ad, email, or social post needs.
Let the canvas adapt. Matching ratios resize cleanly; new ratios intelligently extend the scene instead of stretching it.
+ /900x400_mountainsinbackground
One segment appended to the URL. The square portrait becomes a wide mountain scene β canvas extended, llama intact.
06 Embeddable editor
When your users or content team need to tweak an image in place, mount the Inliner Studio editor in an iframe or with InlinerEditor.embedIframe(). Pass any hosted sourceUrl and receive the saved lossless WebP URL over postMessage.
useContextMask=1) with GPT Image 2.5 Sunburst so unmasked pixels stay untouched..webp (or .png) directly to your project namespace and posts the new CDN URL back to the host window.
{ type: "publish", payload: { url: "https://img.inliner.ai/acme/watch-copper-bezel.webp" } }
07 Creative control
Compare up to four variations side by side, lock reference images per project, or pick a specific generator for the job. Leave routing on Auto for everyday requests, or target a model directly in Studio, URLs, CLI, and MCP:
08 Pricing
Every plan draws from one monthly credit pool across Studio, URLs, API, CLI, and MCP. Cheap Mode uses 1 credit; standard generation uses 4.
One project to prove the URL workflow on something real.
A recurring personal workflow: side projects, a blog, a portfolio.
Small teams and client work with a reusable asset library.
High-volume production assets across many brands and pipelines.
One monthly pool across Studio, URL, API, CLI, and MCP. Reusing the CDN URL is free.
Lower-cost models. Four times more images from the same pool.
Grok, Gemini, GPT Image, Flux, and the rest.
Edits, infill, and enhancement keep their own allowances. Failed generations are refunded. Pools are 4Γ larger, so a standard image still costs the same share of the plan.
Included paid-plan generations land around 6β7Β’ when the monthly allowance is fully used β versus $2β$10 for a typical stock image. Need enterprise volume? Talk to us.
09 FAQ
Add your description to your project base URL: https://img.inliner.ai/your-project/your-description-here. The AI processes the description and returns an image as soon as it finishes generating. Subsequent requests to the same URL are served from the CDN for fast, scalable delivery. You can also append .webp to any existing PNG or JPG URL (for example hero.png.webp) to serve a compressed WebP asset on the fly without spending additional credits.
Yes. Once an image is generated, its URL always returns the same image, provided your account remains in good standing (for paid plans). That makes it safe to embed in websites, emails, and applications without worrying about broken links.
Our current lineup includes FLUX 3 Image (default standard generation), Nano Banana 2.1 (default reference-image generation and conversational editing), GPT Image 2.5 Flare (fast typography, packaging, and UI layouts up to 2048Γ2048), GPT Image 2.5 Sunburst (precision local edits and native rectangular context masks), Grok Imagine 2, Krea 2 Medium, Recraft V4.1, Gemini Flash Lite, Z-Image Turbo, and Qwen Image. Pick one per request, use Cheap Mode (1 credit) for budget-aware routing, or let auto-routing select the best fit. Read the FLUX 3 & Nano Banana 2.1 guide and GPT Image 2.5 guide for full details.
Yes. You can mount the full Inliner Studio editor via an iframe or our JavaScript helper (InlinerEditor.embedIframe()). Pass any existing img.inliner.ai URL via sourceUrl, let users edit in Layers mode or AI mode (with Cheap, Recommended, and GPT 2.5 Sunburst model tiers plus optional context masks), and receive the newly published lossless WebP CDN URL via postMessage. See the Embeddable Editor overview and integration tutorial.
Cheap Mode uses 1 image credit. Standard generation uses 4 credits. Credits are shared across Studio, URL, API, CLI, and MCP. Reusing the resulting CDN URL does not consume another generation credit. Failed generations are refunded. Editing, infill, and enhancement use separate plan allowances.
Yes, you have full rights to use generated images for personal or commercial purposes. However, we cannot guarantee that generated content is entirely unique or free from resemblance to existing copyrighted material.
Standard images typically generate in 4β10 seconds, then cache for instant delivery on every subsequent request. Premium models and reference-image generations may take 20β30 seconds. Once cached, the global CDN serves the asset near-instantly worldwide.
Yes β Studio generates and compares up to four variations at once, and the CLI, SDKs, and MCP server make batch generation scriptable for pipelines and agents.
Write it, ship it, and never re-upload it. Free tier, no credit card.