Crun AI, the unified API that reaches 100-plus image, video, and audio models through one endpoint, added an Infinite Canvas — a node-based workspace where each model is a node you connect into a repeatable content-generation workflow, no integration code required.
2026-08-19 · by Moe Ameen
Crun AI (crun.ai) — a unified AI media API that reaches more than 100 image, video, and audio models through a single OpenAI-compatible endpoint — has added an Infinite Canvas, a visual layer that sits on top of that model catalog. The move takes Crun from a developer API you call in code to a workspace a creator can drive by hand.
On the canvas, each model is a node. You connect a prompt to an image model, feed the result into a video model, pass the clip to an audio or voice model, and lay out variations across an endless workspace — chaining Crun's catalog into a repeatable content-generation pipeline without writing integration code. The same node-graph pattern shows up in other AI canvases; Crun's angle is that the nodes ride its one unified, pay-as-you-go model catalog underneath, so you can compare and swap models (Seedance versus Kling versus Veo on the same prompt) without touching separate provider accounts.
Crun bills per generation on a points/credit system with no mandatory subscription, so cost tracks what you actually render rather than a per-seat fee. Exact canvas features, supported nodes, and pricing move quickly — treat crun.ai and docs.crun.ai as the primary source and confirm the current shape there before relying on specifics.
The launch fits a clear 2026 pattern: the frontier of AI tooling is shifting from "which model" to "how you orchestrate many models," and the visual canvas is becoming the default surface for that orchestration.
A visual canvas removes real friction — no more stitching provider SDKs together to test a text-to-image-to-video pipeline. But it moves the bottleneck rather than removing it. What comes off the canvas is a raw clip, image, or audio track, and the expensive part of a creator's week was never rendering one asset; it was turning that asset into a consistent, on-brand, published presence across every feed. That is the job [Kompozy](/) does, and the two fit together without overlap: orchestrate and render on Crun, then hand the output to Kompozy to brand, format, and ship.
Concretely, drop a finished canvas render into Kompozy and it governs everything the canvas can't. Your [Persona Brief](/glossary/persona-brief) and banned-word rules keep the copy in one voice; Gemini face-lock holds a persona's identity across pieces so a set of posts looks like one creator, not a pile of unrelated experiments; and each asset is [clipped](/glossary/clipped-short) and reframed to 9:16, 1:1, and 16:9 with burned-in captions before it ships across the eight social platforms plus blog and email — on a per-post review gate or full [Autopilot](/glossary/autopilot). Because Kompozy supports bring-your-own-key, you keep Crun's model optionality and still hand off brand voice and publishing. The canvas is where you find the shot; Kompozy is where one shot becomes a governed, scheduled content week — plus the [Persona Shorts](/glossary/persona-shorts) and avatar video a model canvas can't assemble.
It is a visual, node-based workspace Crun AI added on top of its unified model API. Each of Crun's 100+ image, video, and audio models is a node you connect on an endless canvas, chaining them into a repeatable content-generation workflow without writing integration code. Confirm the current feature set on crun.ai.
Crun uses pay-as-you-go, per-generation pricing on a points/credit system with no mandatory subscription, so cost tracks what you actually render. Rates vary by model and move with the underlying providers, so check crun.ai for current per-model pricing.
No. The canvas generates raw assets — clips, images, and audio — and has no brand-voice layer, captioning, per-platform reframing, or scheduler. To turn a render into on-brand posts and publish across platforms, pair it with a content engine like Kompozy, which owns branding, formatting, and multi-platform publishing.