Z-Image Turbo by Alibaba — 6B-param speed model by Tongyi-MAI. 1–8 step inference at $0.005/MP with LoRA and tiling support.
Multiple Z-Image variants for generation, editing, tiling, and ControlNet — all on the same Runflow API.
Speed-optimised — 6B params, 1–8 steps, $0.005/MP
LoRA fine-tuning variant for custom styles
Seamless tiling for textures and patterns
Mask-based inpainting and region editing
Guided generation with structural control
Every model on Runflow is scored per vertical — not generic leaderboard numbers. Scores come from domain-expert evaluation on real production prompts.
| Use Case | Assessment | Notes |
|---|---|---|
| Portrait | Moderate | Fast drafts at very low cost — good for previews and iteration |
| Fashion | Moderate | Rapid concept generation for high-volume lookbook workflows |
| Product | Moderate | Quick product shots for catalogue-scale generation |
| Creative | Strong | Great for tiling, texture generation, and creative exploration |
Full parameter reference for the z-image-turbo model endpoint.
| Parameter | Type | Description |
|---|---|---|
| prompt | string | Text description of the image to generate. |
| image_size | object | string | Output size. Presets: square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9, auto. Or custom {width, height}. Max 4MP. Default: landscape_4_3. |
| num_inference_steps | integer | Number of denoising steps. Range: 1–8. Default: 8. Fewer = faster, more = better detail. |
| acceleration | string | Speed optimisation level. Options: "none", "regular", "high". Default: "regular". |
| num_images | integer | Number of images to generate per request. Range: 1–4. Default: 1. |
| output_format | string | Output format. Options: "jpeg", "png", "webp". Default: "png". |
| enable_safety_checker | boolean | Enable NSFW content filtering. Default: true. |
| enable_prompt_expansion | boolean | Automatically expand the prompt for better results. Adds $0.0025 per request. Default: false. |
| seed | integer | Random seed for reproducible outputs. |
| sync_mode | boolean | Return media as data URI without storing in request history. Default: false. |
Flux AI models are available on many platforms. Here's what makes the Runflow image generation API different.
One of the cheapest production-grade models available — a 1024×1024 image costs $0.005. Generate thousands of images for the cost of one premium model image.
Z-Image Turbo compresses inference to 1–8 steps where standard diffusion models need 20–50. Dramatically faster throughput for high-volume pipelines.
LoRA fine-tuning, seamless tiling for textures, ControlNet for structural guidance, inpainting for region editing, and image-to-image — all as API variants.
The acceleration parameter (none/regular/high) provides an additional speed layer on top of the already-fast turbo architecture.
Every output is scored by Runflow's Sentinel system before delivery, catching low-quality or off-prompt results automatically.
Same API, different strengths. Switch models with one parameter change.
Black Forest Labs
FLUX.1 [dev] is a 12B-parameter text-to-image model by Black Forest Labs. Open-w…
ByteDance
Seedream 4.5 by ByteDance — unified generation and editing in one model. Strong…
ByteDance
Seedream 4 by ByteDance — unified generation and editing model with photorealism…
Nano Banana 2 by Google — fast image generation across 4 resolution tiers (0.5K–…
OpenAI
GPT-Image 1.5 by OpenAI — high-fidelity images with strong prompt adherence. Thr…
Nano Banana Pro by Google — premium image generation with sharp detail and natur…
Get API access in minutes. No GPU setup, no infrastructure to manage. Pair with Sentinel to control the quality of every AI-generated image you ship.