Typical synchronous render completes in ~1-3 s. The engine is native libvips, no browser rendering, which is why there is no cold-start penalty.

What image formats and sizes are supported?

PNG output up to 4096×4096 px, up to 40 layers per template: text, images, shapes, QR codes, barcodes (Code128, EAN-13), gradients, groups.

Is there a free tier?

Yes: 100 renders and 20 templates per month, no credit card, full API access. Paid plans from $15/mo.

How do webhooks work?

POST /render/async with a webhook_url; Rendry calls it with the job result when the render finishes. Poll /render/status/{job_id} as an alternative.

Can I generate images without writing templates by hand?

Yes: pick one of 39 gallery templates and clone it via API, design visually in the editor, or let an AI agent design it over MCP from raw JSON DSL.

An image generation API that behaves like infrastructure

One POST request: template ID plus a flat map of variables. One response: a finished PNG at the exact canvas size, rendered by a native libvips engine in ~1-3 s, no headless Chrome, no cold starts, no surprise crops. Design the template visually or as raw JSON DSL; render it from code, cron jobs, n8n/Make/Zapier or your AI agent.

Start free; 100 renders/month · Documentation

Why teams and agents use it

Sync and async. Sync render answers in ~1-3 s (up to 25 s). Async returns a job id and calls your webhook, built for queues and batch jobs.
Dynamic everything. Text {variables}, dynamic image URLs, QR and barcode content, per-request filenames. One template, any number of variations.
Typed contract. OpenAPI spec at /api/v1/openapi.json, strict DSL validation with did-you-mean errors, explicit 4xx codes your code can branch on.
Not just an API. The same templates open in a visual editor for humans, and in a 24-tool MCP server for AI agents. Three interfaces, one source of truth.

One request, one image

curl -X POST https://rendry.io/api/v1/render/sync \
  -H "Authorization: Bearer sk_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "preset_id": "your-template",
    "dynamic_data": { "title": "Hello", "photo": "https://..." }
  }'
# → 200 OK, {"result_url": "https://rendry.io/....png"}

Ready-made templates for this

OG Image — Blog Post: 1200x630 article cover. Dark violet-cyan mesh + grain, category chip, gradient headline, author strip. Mood: modern engineering blog. Use for: blog posts, changelogs, docs. (1200x630)
Price Tag: Shelf price tag with product name and big price. Use for: retail shelves and markets. (500x300)
Payment Receipt: Payment receipt with itemized rows and totals. Use for: order confirmations. (600x900)
Shipping Label: Shipping label with addresses and barcode. Use for: parcels and logistics. (800x600)
Testimonial Card: Customer testimonial with avatar, quote and rating. Use for: reviews and social proof. (1200x628)
X / Twitter Post Card: X/Twitter 1600x900 text card. Near-black with violet mesh, bold white statement, author strip with avatar. Mood: confident hot take. Use for: quotes, opinions, threads. (1600x900)

Built for AI agents, not just humans

Rendry ships a native remote MCP server (https://rendry.io/api/v1/mcp) with 25 tools. Your agent designs templates from raw JSON DSL, previews them for free, then renders batches. Full control over every pixel stays with the agent, while a human can always open the same template in the visual editor and adjust it. No headless browser, no screenshot hacks, no glue code.

Machine-readable: llms.txt · llms-full.txt · openapi.json · mcp.json

Frequently asked questions

How fast is the API?: Typical synchronous render completes in ~1-3 s. The engine is native libvips, no browser rendering, which is why there is no cold-start penalty.
What image formats and sizes are supported?: PNG output up to 4096×4096 px, up to 40 layers per template: text, images, shapes, QR codes, barcodes (Code128, EAN-13), gradients, groups.
Is there a free tier?: Yes: 100 renders and 20 templates per month, no credit card, full API access. Paid plans from $15/mo.
How do webhooks work?: POST /render/async with a webhook_url; Rendry calls it with the job result when the render finishes. Poll /render/status/{job_id} as an alternative.
Can I generate images without writing templates by hand?: Yes: pick one of 39 gallery templates and clone it via API, design visually in the editor, or let an AI agent design it over MCP from raw JSON DSL.