GPT Image 2.5 Sunburst API Cost: A Reproducible Estimate for Privacy-Aware Workloads
Estimate GPT Image 2.5 Sunburst API costs using explicit token assumptions, monthly workload arithmetic, and practical safeguards for user data and credentials.
Summary · TL;DR
At the supplied rates, a workload using 3,000 input tokens and 400 output tokens per request costs about $0.027 per request before retries, taxes, routing fees, or other provider-specific charges. At 5,000 requests per day, that is approximately $135 per day or $4,109 over a 30.4375-day average month. This estimate is illustrative: actual usage depends on tokenization, retries, cache eligibility, batch processing, and the provider's current pricing.
Key Takeaways
- The supplied GPT Image 2.5 Sunburst rates are $5.00 per 1 million input tokens and $30.00 per 1 million output tokens.
- A 3,000-input-token and 400-output-token request costs $0.015 plus $0.012, or $0.027 before operational overhead.
- At 5,000 such requests per day, the estimated monthly spend is about $4,109 using a 30.4375-day month.
- The supplied source lists cached input at $1.25 per million tokens and batch input at $2.50 per million tokens, but eligibility and applicability must be confirmed before using those rates.
- Do not place passwords, API keys, session tokens, or unredacted personal data in prompts or image inputs merely because the model supports vision or PDF input.
On this page (9)
Short answer
For the illustrative workload below, GPT Image 2.5 Sunburst costs $0.027 per request:
- 3,000 input tokens × $5.00 per million = $0.015
- 400 output tokens × $30.00 per million = $0.012
- Total = $0.027 per request
At 5,000 requests per day, the estimate is $135 per day and approximately $4,109 per 30.4375-day month. This is a planning estimate, not a quote. It excludes retries, failed requests that are still billable, taxes, and any routing or platform charges not included in the provider rates.
OneMux is the publisher of this guide and provides access to models through one OpenAI-compatible API. Its published catalogue does not provide input or output prices for GPT Image 2.5 Sunburst, so this article uses the dated provider pricing supplied by the external source rather than treating missing OneMux values as zero.
Pricing snapshot and billing units
The supplied source identifies GPT Image 2.5 Sunburst as an OpenAI image generation model with vision support. Its rates are expressed in USD per 1 million tokens, not per image or per request. The source was last verified on October 22, 2026.
| Price category | GPT Image 2.5 Sunburst rate | Unit | Condition or limitation |
|---|---|---|---|
| Input | $5.00 | Per 1 million tokens | Standard input rate in the supplied source |
| Output | $30.00 | Per 1 million tokens | Standard output rate in the supplied source |
| Cached input | $1.25 | Per 1 million tokens | Confirm cache eligibility and provider treatment |
| Batch input | $2.50 | Per 1 million tokens | Confirm batch availability and processing conditions |
| OneMux input | Unknown | Not published | Do not treat as free or zero |
| OneMux output | Unknown | Not published | Do not treat as free or zero |
| OneMux discount ratio | Unknown | Not published | No published discount is asserted |
Source: Future AGI GPT Image 2.5 Sunburst pricing and calculator. For current OneMux catalogue information, see the GPT Image 2.5 Sunburst model page.
Reproduce the estimate
Use this calculation for any workload where token counts and billing units match the rates above:
input_cost = input_tokens / 1,000,000 × 5.00
output_cost = output_tokens / 1,000,000 × 30.00
request_cost = input_cost + output_cost
monthly_cost = request_cost × requests_per_day × days_per_month
With the stated assumptions:
input_cost = 3,000 / 1,000,000 × 5.00 = 0.015
output_cost = 400 / 1,000,000 × 30.00 = 0.012
request_cost = 0.015 + 0.012 = 0.027
monthly_cost = 0.027 × 5,000 × 30.4375 = 4,108. ???
Rounded to cents, the monthly result is approximately $4,109. The source calculator reports the same approximate monthly figure using 5,000 requests per day. If your finance team uses a 30-day month instead, the result is $4,050. State the month-length convention whenever comparing estimates.
A more useful sensitivity check
If the request mix changes, output tokens have a larger effect because the output rate is six times the input rate. For example, keeping 5,000 requests per day and a 30-day month:
| Input tokens/request | Output tokens/request | Estimated monthly cost |
|---|---|---|
| 3,000 | 400 | $4,050 |
| 3,000 | 800 | $7,650 |
| 6,000 | 400 | $4,500 |
These are illustrative calculations using the supplied standard rates. They do not measure model behavior or guarantee that a particular prompt will produce those token counts.
Privacy-aware handling of user data and credentials
GPT Image 2.5 Sunburst is documented as supporting vision and PDF input, but that capability is not a reason to send raw secrets. Treat every prompt, uploaded image, PDF, and generated result as potentially sensitive.
A safer request path is:
- Classify the input before sending it. Separate public content, personal data, confidential business data, credentials, and regulated information.
- Reject secrets at the application boundary. Never forward passwords, private keys, API keys, bearer tokens, recovery codes, or full session cookies to an image model.
- Redact rather than merely instruct. Replace account numbers, email addresses, faces, access tokens, and identifying document fields with stable placeholders before tokenization or upload.
- Use least-privilege service credentials. Keep the model API key in a server-side secret store. Do not put it in browser code, prompts, generated images, logs, or client-visible error messages.
- Minimize retention. Record request identifiers, token counts, cost, and policy decisions where needed, but avoid storing raw prompts and images unless there is a documented business reason.
- Review generated output. An output can reproduce sensitive text or visual details from an input. Apply access controls and redaction before returning it to a user or saving it.
The source also says that function calling is not advertised for this model. Do not assume the model can securely retrieve credentials or call internal systems. Keep authentication, authorization, retrieval, and tool execution in your own controlled application layer.
Using OneMux conditionally
OneMux describes its API as OpenAI-compatible and provides unified model access, routing, keys, spend visibility, credit top-ups, and pay-as-you-go usage. Compatibility still depends on the documented endpoint, model identifier, supported image parameters, and response format for your account. This example is illustrative and was not tested here:
from openai import OpenAI
client = OpenAI(
api_key="YOUR_SERVER_SIDE_ONEMUX_KEY",
base_url="YOUR_DOCUMENTED_ONEMUX_OPENAI_COMPATIBLE_ENDPOINT",
)
response = client.chat.completions.create(
model="gpt-image-2-5-sunburst",
messages=[
{
"role": "user",
"content": "Create an image from this redacted description: [USER_CONTENT]"
}
],
)
print(response.choices[0].message.content)
Confirm the correct model name, image-generation operation, input modality, usage fields, and endpoint in the current OneMux documentation or model catalogue. Do not assume a chat-completions example supports image output without checking the documented parameters.
Cost and privacy checklist
Before approving a production workload, verify:
- Input and output counts are measured in the same token units as the rate table.
- The estimate includes retries, moderation rejections, timeouts, and duplicate submissions.
- Cached or batch rates are used only when the provider confirms eligibility.
- OneMux prices are confirmed separately because the published snapshot lists them as unknown.
- Secrets are blocked before tokenization, upload, logging, and error reporting.
- Raw images and prompts have an explicit retention period.
- The application does not rely on undocumented function calling or credential access.
- A small controlled pilot records actual usage before a monthly budget is approved.
Conclusion
For a transparent starting point, budget GPT Image 2.5 Sunburst at $0.027 per request under the 3,000-input/400-output assumption, or about $4,109 per average month at 5,000 requests per day. Recalculate with observed token counts, retry rates, and confirmed cache or batch eligibility. For privacy-sensitive workloads, the most important control is upstream: redact user data and reject credentials before the request reaches the model or an API gateway.
Published price snapshot
Captured 2026-11-17T17:00:06.096Z from the OneMux published catalogue. Prices retain website billing units; this is not a live benchmark. Tiered, cached or time-based billing may differ. Unknown values are not zero.
| Model | Official input | Official output | OneMux input | OneMux output | Published price ratio |
|---|---|---|---|---|---|
| GPT Image 2.5 Sunburst | Unknown | Unknown | Unknown | Unknown | Unknown / not discounted |
Current OneMux prices. Compare workload costs only when input/output and billing units match; see each model page for tier conditions.
FAQ
How much does GPT Image 2.5 Sunburst cost per request?
Using the supplied example of 3,000 input tokens and 400 output tokens, the cost is $0.015 for input plus $0.012 for output, or $0.027 per request before retries and other charges.
What are the GPT Image 2.5 Sunburst input and output prices?
The supplied source lists $5.00 per 1 million input tokens and $30.00 per 1 million output tokens, last verified October 22, 2026.
Does OneMux publish prices for GPT Image 2.5 Sunburst?
The supplied OneMux catalogue snapshot lists both OneMux input and output prices as unknown. Missing values must not be treated as zero, so confirm current pricing before committing to a budget.
Can I send credentials to GPT Image 2.5 Sunburst?
No. Do not send passwords, API keys, private keys, bearer tokens, session cookies, or recovery codes. Reject or redact secrets before prompts, images, PDFs, logs, and uploads reach the model.
Does GPT Image 2.5 Sunburst support vision and PDF input?
The supplied source documents vision and PDF input as supported modalities. It does not advertise function calling, so do not assume it can securely retrieve credentials or invoke internal tools.
Are cached and batch prices guaranteed for every request?
No. The supplied source lists cached input at $1.25 per million tokens and batch input at $2.50 per million tokens, but eligibility, availability, and processing conditions must be confirmed.
Related articles
Guides
Kimi K3.1 Pro Highspeed API Cost Calculator for Privacy-Aware Workloads
Estimate Kimi K3.1 Pro Highspeed API costs with explicit token, cache, and workload assumptions, then design a safer credential-handling workflow.
Guides
Gemini 4 Argon API Cost: A Privacy-Aware Workload Estimate
Estimate Gemini 4 Argon API costs using explicit token assumptions, then account for retries, caching, and the privacy risks of user data and credentials.
Guides
Claude Haiku 5.5 API Cost Estimates for Privacy-Aware Workloads
Estimate Claude Haiku 5.5 API costs with explicit token assumptions, tier-aware arithmetic and practical steps for minimizing exposure of personal data and credentials.
Guides
DeepSeek V4.1 Flash 1017 API Cost: A Reproducible Workload Estimate
Estimate DeepSeek V4.1 Flash 1017 costs from explicit token assumptions, published OneMux and official rates, time tiers, and cache-read pricing—with privacy steps for user data and credentials.