Short answer
OpenAI’s API model page currently labels GPT-5 Pro as deprecated and lists the alias gpt-5-pro and snapshot gpt-5-pro-2025-10-06. The page says GPT-5 Pro is available through the Responses API only. Because the model is deprecated, it is not a sensible default for a new production integration. The model page does not show a shutdown date; OpenAI says to check the model’s shutdown_date through the List models or Retrieve a model API. (GPT-5 Pro model page; Deprecations)
GPT-5.5 Pro is documented separately, with its own model ID, limits, API availability, and prices. OpenAI’s cited material does not identify it as GPT-5 Pro’s official replacement, so evaluate it as a distinct option rather than assuming the names refer to the same model. (GPT-5.5 Pro model page)
What GPT-5 Pro refers to
GPT-5 Pro is an API model—not the ChatGPT Pro subscription. OpenAI’s current model page identifies its snapshot as gpt-5-pro-2025-10-06, marks it deprecated, and lists gpt-5-pro as its alias. Its model-specific guidance says to use the Responses API. OpenAI does not state a replacement recommendation on that page. (GPT-5 Pro model page)
Before planning around an existing GPT-5 Pro integration, check the model’s live shutdown_date field and test access in your API project. The retrieved model page confirms the deprecation status but does not specify the shutdown date. (Deprecations)
GPT-5 Pro capabilities and constraints
OpenAI’s model page lists these specifications for GPT-5 Pro specifically:
- Reasoning:
reasoning.effort: highonly. - Context and output: 400,000-token context window; 272,000 maximum output tokens.
- Knowledge cutoff: September 30, 2024.
- Modalities: text and image input; text output. Audio and video are not supported.
- Features: streaming, function calling, and structured outputs are listed as supported; fine-tuning and predicted outputs are not. Code Interpreter is not supported.
- Latency: OpenAI says some requests may take several minutes and suggests background mode to avoid timeouts.
These are documented specifications, not a guarantee of quality for a particular workload. Since the model is deprecated, weigh its documented fit against the migration and shutdown risks before using it for new production work. (GPT-5 Pro model page)
Production API: endpoint, model ID, and a smoke test
For GPT-5 Pro, the model-specific documentation says to use the Responses API. The alias is gpt-5-pro; the dated snapshot is gpt-5-pro-2025-10-06. As the model is deprecated, check its current shutdown date and availability before running a test or relying on it. (GPT-5 Pro model page; Deprecations)
For an existing integration or a brief compatibility check, the following is a copyable Python example. It uses the documented SDK setup pattern and the GPT-5 Pro alias; it is not a recommendation to start a new production dependency on a deprecated model. (API quickstart)
pip install openai
export OPENAI_API_KEY="your_api_key_here"
python - <<'PY'
from openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-5-pro",
input="Summarize the main risks in this project plan.",
)
print(response.output_text)
PY
If a request may run for minutes, plan for asynchronous handling: OpenAI’s model page recommends background mode for avoiding timeouts on long-running requests. (GPT-5 Pro model page; Background mode)
GPT-5 Pro API pricing and a transparent estimate
OpenAI lists these standard GPT-5 Pro token prices:
| Token type | Price per 1 million tokens |
|---|---|
| Input | $15.00 |
| Output | $120.00 |
Reasoning tokens are billed as output tokens, so an estimate should count total billable output—including reasoning tokens—not just the visible answer. The estimate below assumes 10,000 input tokens and 2,000 total billable output tokens:
- Input: 10,000 ÷ 1,000,000 × $15 = $0.15
- Output: 2,000 ÷ 1,000,000 × $120 = $0.24
- Estimated total: $0.39 per request
This is an arithmetic illustration, not a typical-cost prediction. Actual usage depends on the complete input and output token counts. The GPT-5 Pro model page lists the standard rates and a 400,000-token context window, but the retrieved model-page text does not specify a long-context price threshold or a regional-processing uplift for GPT-5 Pro. Check current pricing for your actual request pattern rather than assuming those conditions. (GPT-5 Pro model page; Reasoning-token billing)
What about GPT-5.5 Pro?
GPT-5.5 Pro is a separately documented model, not an established alias or official replacement for GPT-5 Pro. Its model page says it is available through Responses API requests, including Batch API requests, and lists gpt-5.5-pro and snapshot gpt-5.5-pro-2026-04-23. Its documented context window is 1,050,000 tokens, maximum output is 128,000 tokens, and reasoning effort supports medium, high (the default), and xhigh. Its listed standard prices are $30 per million input tokens and $180 per million output tokens. (GPT-5.5 Pro model page)
The GPT-5.5 Pro page also states that it offers no cached-input discount and that regional-processing (data-residency) endpoints carry a 10% uplift. The retrieved page text does not specify a GPT-5.5 Pro long-context price threshold; verify current pricing if your workload approaches its context limit. Do not apply GPT-5.5 Pro’s limits or billing terms to GPT-5 Pro. (GPT-5.5 Pro model page)
Bottom line
If you searched for GPT-5 Pro, the key implementation fact is that OpenAI currently marks gpt-5-pro deprecated. Its model page documents Responses API access, a high-only reasoning setting, a 400,000-token context window, 272,000 maximum output tokens, and standard rates of $15/$120 per million input/output tokens. Check the model’s live shutdown date before relying on it; the captured documentation does not establish an official replacement. Treat GPT-5.5 Pro as a separate option to evaluate, not as an assumed successor.
Sources
GPT-5.5 Pro Model | OpenAI API