Alibaba's open-source 20B MMDiT with superior text rendering, photorealistic humans, and tunable acceleration — the largest open-weight image model at $0.02/megapixel.
Explore all variants in the Qwen Image generation family.
Latest 20B model with tunable acceleration
Customizable variant with LoRA support
Previous generation 7B model
Pro variant of the previous generation
Premium quality variant
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 | Pending | Portrait quality |
| Fashion | Pending | Fashion quality |
| Product | Pending | Product quality |
| Creative | Pending | Creative quality |
Full parameter reference for the qwen-image-2512 model endpoint.
| Parameter | Type | Description |
|---|---|---|
| prompt | string | Text description of the image to generate. Supports both English and Chinese prompts with strong multilingual understanding. |
| negative_prompt | string | Describe what to exclude from the generated image — artifacts, styles, objects, or quality issues. |
| acceleration | string | Speed/quality tradeoff. 'none' for maximum quality, 'regular' for balanced, 'high' for fastest generation. |
| image_size | object | string | Output dimensions. Use preset strings or specify {width, height}. Default: landscape_4_3. |
| num_images | integer | Number of images to generate per request. Range: 1–4. |
| num_inference_steps | integer | Number of denoising steps. 20 for fast drafts, 28 for balanced, 35–50 for max fidelity. Default: 28. |
| guidance_scale | number | Controls prompt adherence. Lower values (2–4) more creative; higher (5–8) more strict. Default: 4. |
| seed | integer | Reproducibility seed. Same seed + same parameters = identical output. |
| output_format | string | Output image format. Options: "jpeg", "png", "webp". Default: "png". |
| enable_safety_checker | boolean | If set to true, the safety checker will be enabled. |
| sync_mode | boolean | If true, returns the image as a data URI. |
Flux AI models are available on many platforms. Here's what makes the Runflow image generation API different.
Choose between three acceleration modes (none, regular, high) to balance quality vs speed and cost per your use case.
Major improvement over Qwen Image 2.0 in rendering readable text within images — handles Chinese, English, and mixed-language typography.
Fine-grained control over skin textures, facial features, and body proportions with fewer artifacts than previous generations.
At $0.02/megapixel, Qwen Image 2512 is one of the most affordable high-quality image generation models available via API.
Improved rendering of materials like fabric, metal, wood, and skin — producing outputs that need minimal post-processing.
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.