> ## Documentation Index
> Fetch the complete documentation index at: https://docs.laozhang.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Legacy Grok Imagine image API

> Legacy Grok Imagine models grok-imagine-image and grok-imagine-image-quality, launched August 13, 2026: ratios, 1K/2K, reference limits, and pricing.

`grok-imagine-image` and `grok-imagine-image-quality` are the legacy Grok Imagine image models on LaoZhang API. They launched on August 13, 2026 under the label “Grok Imagine 2”, remain available through the Images API, and are billed per generated image.

* **Published**: August 13, 2026
* **Last verified**: September 24, 2026
* **Status**: Active (legacy model)

## Key facts

| Item | Detail |
| - | - |
| Standard model | `grok-imagine-image` |
| Quality model | `grok-imagine-image-quality` |
| Newer model (separate) | `grok-imagine-image-2.0`, launched September 3, 2026; see the [Grok Imagine Image 2.0 launch notice](/en/announcements/grok-imagine-image-2-0-2026-09) and the [Grok Imagine Image API guide](/en/api-capabilities/grok-imagine-image) |
| Aspect ratios | `1:1`, `16:9`, `9:16`, `4:3`, `3:4` |
| Resolution | `1k` and `2k`; same price within each model |
| Batch generation | `n` up to 10, billed by generated image count |
| Standard reference editing | `/v1/images/edits`; 1–4 references |
| Quality reference editing | `/v1/images/edits`; 1–2 references |
| Text-to-image | `/v1/images/generations` |
| Response format | `url` or `b64_json` |

xAI documents more aspect ratios and uses its own price list. On LaoZhang API, the legacy models support the five ratios above, and prices are on the [LaoZhang API console](https://api.laozhang.ai/account/pricing) and in call logs.

“Up to 10 images” is the output count set by `n`, not a reference-input limit. Reference limits are model-specific: Standard accepts 1–4 references and Quality accepts 1–2.

## Pricing

Public pricing as of September 24, 2026, per successfully generated image:

* `grok-imagine-image`: \$0.025. It launched at \$0.02 and changed on September 3, 2026.
* `grok-imagine-image-quality`: \$0.045, unchanged since launch.

Current prices are in the [model and pricing catalog](/en/models). Use console call logs as the billing record.

## Who should care

* Existing integrations that already use these model IDs and need predictable per-image budgets for 1K/2K batch generation;
* Content, advertising, and commerce teams that want multiple candidates in one request;
* Multi-reference workflows that need four references: `grok-imagine-image` accepts up to four, while `grok-imagine-image-2.0` accepts up to three;
* Teams migrating from GPT Image, FLUX.2, or another Images API that want OpenAI SDK compatibility.

Text, voice, and video calls are not affected. Workflows that require 4K, mask inpainting, arbitrary pixel dimensions, or reproducible `seed` behavior should not assume direct compatibility.

## What to do

<Steps>
  <Step title="Confirm token group and price">
    Open [console model pricing](https://api.laozhang.ai/account/pricing) and verify that the target token can see the model ID and its current rate.
  </Step>

  <Step title="Choose standard or quality">
    Use `grok-imagine-image` for routine assets and three- or four-reference composition. Compare `grok-imagine-image-quality` for final assets or one- to two-reference precision edits.
  </Step>

  <Step title="Test one image first">
    Test the target ratio and resolution with `n=1`. Inspect actual dimensions, image content, response fields, and the charge before increasing batch size or concurrency.
  </Step>
</Steps>

Minimal text-to-image request. Replace the placeholder with your LaoZhang API key; the download step uses `jq`:

```bash theme={null}
export LAOZHANG_API_KEY="YOUR_LAOZHANG_API_KEY"

curl -sS https://api.laozhang.ai/v1/images/generations \
  -H "Authorization: Bearer $LAOZHANG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine-image",
    "prompt": "A minimalist coffee product poster with warm studio lighting",
    "n": 1,
    "aspect_ratio": "1:1",
    "resolution": "1k",
    "response_format": "url"
  }' \
  -o grok-legacy.json

curl -sS -o grok-legacy.jpg "$(jq -r '.data[0].url' grok-legacy.json)"
```

The response lists the image URL in `data[0].url`. The URL is temporary, so download it promptly. For Python, reference editing, multi-image composition, migration, and errors, see the [Grok Imagine Image API guide](/en/api-capabilities/grok-imagine-image).

## Frequently asked questions

### What are the correct model IDs for the legacy models?

Use `grok-imagine-image` and `grok-imagine-image-quality`. The launch label “Grok Imagine 2” is not part of either ID, and it does not refer to `grok-imagine-image-2.0`, which is a separate, newer model.

### Is pricing per request or per image?

Per successfully generated image, so `n=10` can create ten image charges. The per-image rates are in the Pricing section above; use the console call log as the billing record.

### Did the price change after launch?

Yes. `grok-imagine-image` changed on September 3, 2026; the Pricing section lists the launch and current rates. `grok-imagine-image-quality` did not change.

### Do 1K and 2K cost the same?

Yes, within each model under current LaoZhang pricing. Pricing may change, so recheck the console before a large batch.

### Why does this page list only five aspect ratios?

They are the documented LaoZhang support scope for these models. xAI documents more upstream ratios, but upstream capability alone does not prove that every value works on the LaoZhang gateway.

### Which endpoint accepts reference images?

Use `/v1/images/edits` with `multipart/form-data`:

* Fields: `image` for one file, and repeated `image[]` for multi-image edits.
* Limits: `grok-imagine-image` supports 1–4 references, and `grok-imagine-image-quality` supports 1–2.

Text-to-image continues to use JSON at `/v1/images/generations`.

### Should I migrate to `grok-imagine-image-2.0`?

Not necessarily. The legacy models remain available. Compare them with `grok-imagine-image-2.0` on matched prompts: 2.0 adds 15 fixed aspect ratios and Low/Medium quality but accepts at most three references, while `grok-imagine-image` accepts up to four.

## Evidence and related documentation

* [xAI multi-image editing](https://docs.x.ai/developers/model-capabilities/images/multi-image-editing) — upstream multi-image workflow and ordering
* [xAI Grok Imagine Image model](https://docs.x.ai/developers/models/grok-imagine-image) — standard-model image input
* [xAI image generation documentation](https://docs.x.ai/developers/model-capabilities/images/generation) — upstream parameters and response formats
* [xAI models and pricing](https://docs.x.ai/developers/pricing) — xAI list pricing, not LaoZhang API retail pricing
* [Grok Imagine Image 2.0 launch notice](/en/announcements/grok-imagine-image-2-0-2026-09) — the newer model, endpoints, and pricing
* [Grok Imagine Image API guide](/en/api-capabilities/grok-imagine-image) — LaoZhang endpoints, code, and errors
* [Image generation API selection guide](/en/api-capabilities/image-generation-guide) — compare image models
* [LaoZhang API console](https://api.laozhang.ai/account/pricing) — current groups, prices, and recorded charges


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.