Direct answer: Pick Stable Image Core for speed and low cost, SD3 for balanced quality and prompt adherence, SDXL for a mature ecosystem of fine-tunes and ControlNet, and Stable Image Ultra for final high-quality assets. Older SD 1.5 still works for lightweight, heavily fine-tuned tasks.

What actually differs between the models?

Three axes: image quality, prompt adherence, and cost. Newer models generally look better and follow instructions more closely, but cost more per image and, in some cases, run slower. Older models like SD 1.5 have enormous community tooling — LoRAs, ControlNet adapters, fine-tunes — that newer releases do not match yet.

How do you pick in practice?

1. Start from the output. Is this a throwaway concept or the final asset? Concepts go to Core; finals go to Ultra.
2. Check prompt complexity. Simple single-subject prompts work fine on cheaper models. Multi-subject, spatial, or text-in-image needs SD3 or Ultra.
3. Consider tooling needs. If you rely on specific LoRAs or ControlNet, SDXL and SD 1.5 have the deepest ecosystem.
4. Factor latency. Interactive apps need fast models; batch overnight jobs can afford slower, higher-quality ones.
5. Run a 20-prompt test. Send the same prompts through two candidates and compare side by side. That beats reading model cards.

Which model for which job?

Does the newest model always win?

No. Newer models can be worse for a specific fine-tuned style you already built on SDXL. If your pipeline depends on a particular checkpoint or LoRA, upgrading means rebuilding that work. Stick with what your tooling supports unless the quality gap genuinely matters for your output.

What is often overlooked?

Cost is per generation, and iteration is the real expense. A cheaper model that needs three attempts can cost more than an expensive one that lands the first time. Measure cost per usable image, not cost per call. Teams often optimize the wrong number.

Quick Summary

• Core = cheap and fast; Ultra = high quality; SD3 = balanced; SDXL/SD1.5 = deep tooling.
• Choose by output intent, prompt complexity, latency needs, and existing tooling.
• Newer is not always better if you rely on a specific fine-tune or LoRA.
• Run a 20-prompt head-to-head test rather than trusting model cards.
• Optimize cost per usable image, not cost per API call.

Direct answer: Stability AI’s API generates and edits images from text prompts using models like Stable Image Core, Ultra, and SD3. You sign up for credits at platform.stability.ai, get an API key, and call the generation endpoint. Pricing is credit-based, and the Core model is the cheapest per image.

Which Stability AI model should you use?

There are tiers. Stable Image Core is fast and cheap, good for concepting and bulk generation. Ultra is higher quality and slower, for final assets. SD3 sits in the middle with strong prompt adherence. Text-to-image, image-to-image, upscaling, and editing are separate endpoints, so pick the one matching your task rather than forcing everything through text-to-image.

How do you make a first API call?

1. Create an account at platform.stability.ai and top up credits.
2. Generate an API key from your account page and store it as an environment variable.
3. Choose a model. Start with Core while prototyping to keep costs low.
4. Send a request with your prompt, aspect ratio, and an output format. The API returns image data you save or stream to the user.
5. Add a negative prompt for artifacts and unwanted styles — it meaningfully improves output.
6. Handle moderation. Prompts and outputs pass through filters; build a response path for rejected generations.

How much does it cost?

It is credit-based. Credits are purchased in packs, and each generation consumes credits at a rate depending on the model — Ultra costs more per image than Core, and upscaling or editing has its own rate. There is no free unlimited tier; you buy credits and spend them. Test with Core and small batches before committing to a production budget.

Should you use the API or self-host?

It depends on your team. The hosted API skips GPU management and scales on demand, but you pay per call and depend on their uptime. Running Stable Diffusion yourself needs a GPU and maintenance but has no per-image cost once hardware is in place. High, steady volume favors self-hosting; spiky or low volume favors the API.

Where does it fall short?

Prompt adherence for complex scenes is not perfect — multi-subject compositions still need iteration. Text rendering inside images remains unreliable across the category. And the moderation filters can reject legitimate prompts, which is a support issue in consumer apps.

Quick Summary

• Stability AI’s API generates and edits images via models like Core, Ultra, and SD3.
• Sign up for credits, get a key, and call generation endpoints; Core is cheapest.
• Pricing is credit-based, no free unlimited tier — test with Core first.
• Hosted API skips GPU work; self-hosting suits high steady volume.
• Complex scenes and text-in-image still need iteration and can hit moderation filters.