Add AI generation to your app

Generate — Z-Image-Turbo; Enhance — Llama 3.1 8B Prompt Enhancer; Describe — Qwen2.5-VL-7B-Instruct. Each row above expands with the model's live pricing, capability chips, and sample outputs.

Z-Image-Turbo and Llama 3.1 8B Prompt Enhancer and Qwen2.5-VL-7B-Instruct run on Runware's own optimized compute (the platform's open-weight tier, billed on compute time) — check each model page for weight availability and license terms before self-hosting. The remaining picks are partner-served.

A prompt-enhancement stage between user input and generation lifts output quality, but the only model in that collection today (Llama 3.1 8B Prompt Enhancer) publishes no per-call price. Verify its cost before putting it in front of every generation, and design the flow to degrade to the raw prompt.

At published rates a generate-plus-caption round trip is roughly $0.0032 per user action at 512x512. Multiply by your daily action count before choosing default resolutions; area-based pricing means a 2048x2048 default is ten times the cost.

Every price on this page is the model's published rate from the live Runware catalog, using the cheapest listed configuration unless stated otherwise. Prices vary with resolution, duration, quality tier, or token volume, so check the pricing table on each model page before estimating unit economics.

The Runware catalog does not publish per-model latency figures, so this page does not quote end-to-end timings. Where a model's own description commits to speed (for example sub-second generation or realtime streaming), that claim is repeated here. For anything else, benchmark the exact models in the Playground with your own payload sizes before committing to an SLA.

Models are addressed by versioned AIR identifiers, so a workflow pinned to specific model versions keeps producing the same behaviour as new versions ship. Adopt upgrades deliberately by re-running your evaluation set against the new version before switching production traffic.

A blueprint bundles the small set of models you would actually wire together for one production use case, instead of ranking a whole category. Each pick covers one stage of the workflow and links to its model page for full schema, pricing, and examples.

Picks are live models from the Fastest Image Generation (17 models), Best Prompt Enhance (1), and Best Captioning (5) collections, selected for the lowest published per-call cost in each stage so the stack can sit inside an existing product's unit economics.