Wavemaker Blog
Page 4 of 12
RSS feed-
Flux, SDXL, and Krea LoRAs: What Runs on Wavemaker Today
An honest map of which LoRA base models train and execute on Wavemaker (krea2, ltx, wan, hunyuan, elevenlabs) versus library-only Flux and SDXL imports.
-
ComfyUI Custom Nodes and Dependency Hell (Why We Don't Execute Them Online)
Custom nodes are arbitrary Python, version pins, and broken graphs — dependency hell that makes multi-tenant Comfy hosting fragile. Wavemaker transpiles supported intent instead of running node code.
-
Idempotency Keys and Webhooks for Generative Media APIs
Why POST retries must not double-charge credits, how Wavemaker's Idempotency-Key and signed kernel_run webhooks work, and patterns for batch jobs, approval gates, and training jobs.
-
Publish Your AI Workflow as an App: From Private Graph to /w/{slug}
Publishing compiles your WorkflowSpec, freezes a version, runs moderation, and lists a Hub app at /w/{slug} with JSON Schema inputs, estimates, remix lineage, and Open or Premium economics.