> Markdown mirror of https://router.one/blog/gpt-image-2-6-api-guide for AI assistants and crawlers. Router One is a unified, OpenAI-compatible LLM API gateway.
> Published: 2026-11-10 · Author: Router One Team

# GPT Image 2.6 API Guide: Flare, Sunburst and Billing

_GPT Image 2.6 Flare and Sunburst on Router One: the two model IDs, the POST /v1/images/generations call, per-image wallet billing, and how to choose._

As of the 2026-11-05 catalog, Router One lists GPT Image 2.6 as two model ids, `gpt-image-2.6-flare` and `gpt-image-2.6-sunburst`, and the 2026-11-10 catalog still lists both. Send either id as `model` to `POST https://api.router.one/v1/images/generations` with a Router One key: the request follows the OpenAI Images API shape, like every image model on the gateway. As of the 2026-11-10 catalog, each id is $0.10 per image, and image calls are charged to the wallet: no tier of the 2026-11-10 plan response lists an image model.

This guide covers the two ids, what OpenAI documents about them, how Flare and Sunburst differ in OpenAI's own words, how they relate to GPT Image 2.5 and gpt-image-2, how billing works, the first request with curl and the OpenAI Python SDK, and the errors the endpoint returns. Catalog, plan and OpenAI statements carry the day they were read; the live catalog is the source of truth for a new request.

## GPT Image 2.6 at a glance

| Field | What the catalog lists (2026-11-10) |
| --- | --- |
| Model ids | `gpt-image-2.6-flare`, `gpt-image-2.6-sunburst` |
| First seen | the 2026-11-05 catalog (the 16:05 UTC read); the catalog gives no release date |
| Input / output | text in, image out |
| Endpoint | `POST /v1/images/generations` (OpenAI Images API shape) |
| Price | $0.10 per image for each id (2026-11-10 catalog) |
| Plans | no tier of the 2026-11-10 plan response lists an image model, so every image is charged to the wallet |
| Model pages | [GPT Image 2.6 Flare](https://router.one/models/gpt-image-2-6-flare), [GPT Image 2.6 Sunburst](https://router.one/models/gpt-image-2-6-sunburst) |

## What OpenAI documents for GPT Image 2.6

OpenAI's [API changelog](https://developers.openai.com/api/docs/changelog) (November 4, 2026 entry; checked 2026-11-10) records the release of GPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare for image generation and editing, with adjustable complexity, SVG generation and animated GIF generation, and the addition of markup-guided editing for supported GPT Image 2.5 and 2.6 models. OpenAI's model pages (checked 2026-11-10) describe the two models this way:

- **[GPT-Image-2.6 Flare](https://developers.openai.com/api/docs/models/gpt-image-2.6-flare)**: image generation and editing for fast iteration; OpenAI suggests it for interactive image generation and editing workflows.
- **[GPT-Image-2.6 Sunburst](https://developers.openai.com/api/docs/models/gpt-image-2.6-sunburst)**: image generation and editing for detailed visual work; OpenAI suggests it for detailed compositions and image editing workflows.

OpenAI's [image generation guide](https://developers.openai.com/api/docs/guides/image-generation) (checked 2026-11-10) adds that, on OpenAI's own Image API, both models take a `complexity` field (`low`, `medium` or `high`, and `high` when it is omitted) and produce SVG and animated GIF output, and it lists `low`, `medium`, `high`, `xhigh` and `max` as the quality options for GPT Image 2.6 and 2.5 Sunburst and Flare.

Those are OpenAI's statements about its own API. Router One is responsible for the catalog entry and for how the gateway serves and bills it: the 2026-11-10 catalog lists text input and image output for both ids, their model pages list one endpoint, `POST /v1/images/generations`, and Router One's [Images API reference](https://router.one/docs/images/createImageGeneration) documents `model`, `prompt`, `n`, `size`, `quality` and `response_format` for that endpoint. Per the reference, Router One passes `size` and `quality` to the model as sent without validating them, and a value the model rejects returns a 400. This guide covers those fields only.

## Flare or Sunburst?

OpenAI gives the difference one line each: Flare for fast iteration, Sunburst for detailed visual work (model pages and image generation guide, checked 2026-11-10). On Router One the two ids take the same request and, as of the 2026-11-10 catalog, the same $0.10 per image, so the choice is about the output, not the budget:

- Use **Flare** while you iterate on a prompt, then send the prompts you shortlist to **Sunburst** and keep whichever output your reviewers prefer.
- Switching is one string: the same key, the same endpoint, a different `model`.
- Dashboard → Logs records each request the gateway sends to a model, with its model, the wallet charge it used, latency and status ([per-request observability](https://router.one/llm-observability)), so the requests for both ids sit side by side.

## GPT Image 2.6, GPT Image 2.5 and gpt-image-2

- **Release order.** Per OpenAI's changelog (checked 2026-11-10), GPT Image 2.5 Sunburst and Flare were released on September 8, 2026, and GPT Image 2.6 Sunburst and Flare on November 4, 2026. OpenAI's image generation guide lists `gpt-image-2` among its earlier GPT Image models.
- **On Router One.** As of the 2026-11-10 catalog, all five GPT Image ids are listed (`gpt-image-2`, `gpt-image-2.5-flare`, `gpt-image-2.5-sunburst`, `gpt-image-2.6-flare` and `gpt-image-2.6-sunburst`), each on `POST /v1/images/generations` and each at $0.10 per image.
- **Moving from 2.5.** On the wire the switch is the model string: `gpt-image-2.5-flare` becomes `gpt-image-2.6-flare`, and `gpt-image-2.5-sunburst` becomes `gpt-image-2.6-sunburst`. The 2026-11-10 catalog still lists both 2.5 ids; run your own prompts on both versions before you move a workload.

## How billing works

- **Per image.** As of the 2026-11-10 catalog, `gpt-image-2.6-flare` and `gpt-image-2.6-sunburst` are $0.10 per image each, with no token accounting on the output. Each generated image is one unit: `n: 2` asks for two images, and you are billed for the images the response's `data` array holds ([Images API reference](https://router.one/docs/images/createImageGeneration)).
- **Plans.** As of the 2026-11-10 plan response, no tier of Router One's Pro, Max or Ultra plans lists an image model, so GPT Image 2.6 calls are charged to the wallet, plan or no plan, and need a wallet balance.
- **Per-key caps.** Give an image pipeline its own key with a `maxSpend` cap ([per-key cost tracking](https://router.one/llm-cost-tracking)). A key's spend cap (maxSpend) limits what that key can charge to your wallet: once its cumulative wallet charges reach the cap, its wallet-billed requests return HTTP 402. Requests covered by your plan's quota draw from the account's shared plan quota and do not count toward any key's cap.
- **Records.** Dashboard → Logs shows the model, the wallet charge, latency and status of each image request the gateway sends to a model; a request refused before that, such as a 401 or a 402, leaves no record.

The model pages show the live price: [GPT Image 2.6 Flare](https://router.one/models/gpt-image-2-6-flare) and [GPT Image 2.6 Sunburst](https://router.one/models/gpt-image-2-6-sunburst).

## Send the first request

1. **Create a key.** Dashboard → API Keys → Create Key. Keys look like `sk-...`.
2. **Top up the wallet.** Top up with a card or Alipay through one hosted checkout, or with USDT/USDC on six chains (Tron, BSC, Ethereum, Polygon, Base, Arbitrum). No US credit card required.
3. **Call the endpoint** with either id in `model`.

```bash
export ROUTER_ONE_KEY=sk-your-router-one-key

curl -X POST https://api.router.one/v1/images/generations \
  -H "Authorization: Bearer $ROUTER_ONE_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-image-2.6-flare",
       "prompt": "Flat-lay product photo of a ceramic teapot, soft morning light"}'
```

With the official OpenAI Python SDK, point `base_url` at the gateway:

```python
import base64, httpx
from openai import OpenAI

client = OpenAI(base_url="https://api.router.one/v1", api_key="sk-your-router-one-key")

result = client.images.generate(
    model="gpt-image-2.6-sunburst",
    prompt="Poster of a lighthouse at dusk, flat colors, no text",
    n=2,
)

for i, item in enumerate(result.data):
    png = base64.b64decode(item.b64_json) if item.b64_json else httpx.get(item.url).content
    with open(f"poster-{i}.png", "wb") as f:
        f.write(png)
```

Router One's reference documents two return forms: `url` (the default; the link expires after about an hour) and `b64_json` (set `response_format`). Handle both, as above, and save the bytes right away. The call is synchronous and returns when generation finishes, so give the client a generous timeout rather than a tight one.

## Common errors

The endpoint returns the gateway's standard error body (`message`, `type`, `code` and `request_id` under `error`); the [error codes page](https://router.one/llm-api-error-codes) has the full list. The ones image requests meet most:

| Status | Code | What to do |
| --- | --- | --- |
| 400 | `INVALID_REQUEST` | A wrong or unlisted model id (`requested model is not available`), a missing prompt, an invalid field, or a `size` or `quality` value the model rejects (`invalid request (upstream rejected with status 400)`). Copy the id from [/models](https://router.one/models) and drop the value the model rejects. |
| 400 | `CONTENT_FILTERED` | The prompt failed content moderation (`prompt rejected by content moderation`). No model is called and nothing is billed; change the prompt instead of retrying automatically. |
| 401 | `AUTH_INVALID_API_KEY` | The key is missing or wrong: send it as `Authorization: Bearer sk-…`. |
| 402 | `INSUFFICIENT_BALANCE` or `API_KEY_SPEND_CAP_EXCEEDED` | The wallet balance is used up, or the key reached its `maxSpend`. A top-up does not raise a key's cap. |
| 429 | `RATE_LIMIT_EXCEEDED` | Back off and retry with your own backoff: per the reference, image endpoints do not return the rate-limit headers or `Retry-After`. |
| 502 | `MODERATION_UNAVAILABLE` | Moderation could not run, so the request was refused rather than generated unchecked; retry shortly. |
| 503 | `PROVIDER_UNAVAILABLE` | The upstream is unavailable; retry later. |

## FAQ

**What are the GPT Image 2.6 model IDs on Router One?**
`gpt-image-2.6-flare` and `gpt-image-2.6-sunburst`, both listed as of the 2026-11-10 catalog and first seen in the 2026-11-05 catalog. Put either one in `model` on `POST /v1/images/generations`.

**How much does GPT Image 2.6 cost on Router One?**
As of the 2026-11-10 catalog, each id is $0.10 per image, charged to the wallet. The live price is on each model page.

**Do subscription plans cover GPT Image 2.6?**
No. As of the 2026-11-10 plan response, no tier of the Pro, Max or Ultra plans lists an image model, so every GPT Image 2.6 image is charged to the wallet.

**Should I use Flare or Sunburst?**
OpenAI describes Flare as its GPT Image 2.6 model for fast iteration and Sunburst as the one for detailed visual work (checked 2026-11-10). On Router One both take the same request at the same price, so run your own prompts through both and compare the images.

**Can I switch from GPT Image 2.5 by changing the model ID?**
Yes, the request stays the same: replace `gpt-image-2.5-flare` with `gpt-image-2.6-flare`, or `gpt-image-2.5-sunburst` with `gpt-image-2.6-sunburst`. The 2026-11-10 catalog still lists both 2.5 ids.

**Can I edit a reference image with GPT Image 2.6 on Router One?**
OpenAI's model pages list image editing on OpenAI's own API. Router One's 2026-11-10 catalog lists text input for both 2.6 ids. For reference-image edits on `POST /v1/images/edits`, pick an image model whose catalog entry lists image input, such as `gemini-nano-banana-2.1`; the [image generation API page](https://router.one/image-generation-api) shows which models do.

**Does it work from mainland China?**
Yes. The endpoint is reachable without a VPN, and the wallet takes a card or Alipay on one hosted checkout, or USDT/USDC on six chains.

## Conclusion

For anyone already on the gateway, GPT Image 2.6 is a model string: put `gpt-image-2.6-flare` or `gpt-image-2.6-sunburst` in `model`, send a prompt to `POST /v1/images/generations`, and save whichever of `b64_json` or `url` comes back. As of the 2026-11-10 catalog, both ids are $0.10 per image, charged to the wallet. The [image generation API page](https://router.one/image-generation-api) has the whole image lineup, and the [China image generation guide](https://router.one/blog/image-generation-api-china-guide) covers access and payment from China.

## See also

- Canonical page: https://router.one/blog/gpt-image-2-6-api-guide
- Image generation API: https://router.one/image-generation-api
- LLM API Payment: https://router.one/wechat-pay-llm-api
- All blog posts: https://router.one/blog
- Models and per-model token rates: https://router.one/models (markdown: https://router.one/models.md)
- Pricing: https://router.one/pricing
- API docs (markdown): https://router.one/docs.md
- Company facts: https://router.one/facts/company.md