As of July 23, 2026, GPT-5.2 Pro is no longer OpenAI’s most powerful professional model. OpenAI now describes it as a “previous pro model.” It was removed from ChatGPT on June 12, 2026, along with GPT-5.2 Instant and GPT-5.2 Thinking. However, the official API model page still lists gpt-5.2-プロ for use through the Responses API.
The practical answer is simple: no current ChatGPT subscription plan includes GPT-5.2 Pro, but API users can still choose it for compatibility or reproducibility. For new professional workloads, OpenAI recommends newer models. GPT-5.5 Pro is the direct dedicated-Pro successor, while GPT-5.6 is the current flagship family and provides Pro capability through a reasoning mode rather than a separate Pro model name.
In this guide:
- Current status at a glance
- Definition and specifications
- Availability and plan access
- API価格
- How to use the API model
- Capabilities and limitations
- Successor models
- トラブルシューティング
- Decision guide
- よくあるご質問
GPT-5.2 Pro status at a glance
| 質問 | Current answer on July 23, 2026 |
|---|---|
| Is GPT-5.2 Pro in ChatGPT? | No. OpenAI retired every GPT-5.2 variant from ChatGPT on June 12, 2026. |
| Is GPT-5.2 Pro still in the API? | Yes. OpenAI still lists gpt-5.2-プロ for Responses API use. |
| Does a ChatGPT plan include it? | No current Free, Go, Plus, Pro, Business, Enterprise, or Edu plan provides GPT-5.2 Pro in ChatGPT. |
| Is it still the best OpenAI model? | No. OpenAI’s current flagship API family is GPT-5.6. |
| What replaced it? | GPT-5.5 Pro is the direct dedicated-Pro successor. GPT-5.6 now delivers Pro capability through reasoning.mode: "pro". |
| When does GPT-5.2 Pro still make sense? | Mostly for compatibility, regression testing, or a workflow already validated on its fixed snapshot. |
What is GPT-5.2 Pro?
GPT-5.2 Pro is an older quality-first OpenAI reasoning model for complex professional work. It spends more compute on difficult requests than the standard GPT-5.2 model, accepts text and image input, and returns text. Its main trade-off is unusually high latency and cost compared with newer options.
OpenAI’s current model page publishes the following limits and capabilities:
| 仕様 | GPT-5.2 Pro | What it means in practice |
|---|---|---|
| APIモデル名 | gpt-5.2-プロ | Use this model identifier in a Responses API request. |
| コンテキストウィンドウ | 40万トークン | Large prompts and prior turns must fit inside this total context budget. |
| 最大出力 | 128,000トークン | This is an upper technical limit, not a promise that every response will be that long. |
| 知識カットオフ | 2025年8月31日 | Later facts require current sources, retrieval, or connected tools. |
| 推論努力 | Medium, high, xhigh | Higher effort can improve difficult-task performance but generally increases latency and usage. |
| モダリティ | テキストおよび画像の入力；テキストの出力 | It can analyze images but does not natively return image, audio, or video output. |
| Supported features | Streaming and function calling | It can stream text and call tools through supported API workflows. |
| Unsupported features | Structured outputs, fine-tuning, predicted outputs | Do not choose it when one of these features is a hard requirement. |
Where can you access GPT-5.2 Pro now?
ChatGPT: no current access
GPT-5.2 Pro is not a hidden legacy option in the current ChatGPT model picker. OpenAI’s release notes say GPT-5.2 Instant, Thinking, and Pro are no longer available in ChatGPT, and old GPT-5.2 conversations automatically continue on the corresponding GPT-5.5 model. This retirement applies to ChatGPT, so upgrading to a higher ChatGPT plan does not restore GPT-5.2 Pro.
OpenAI API and Playground: still listed
OpenAI still provides a “Try in Playground” route on the GPT-5.2 Pro model page and documents the model for the Responses API. API billing is separate from a ChatGPT subscription. Access also depends on your API project, billing status, usage tier, rate limits, and OpenAI account controls; the official rate-limit table does not list Free-tier support for this model.
How much does GPT-5.2 Pro cost?
As of July 23, 2026, according to OpenAI’s official GPT-5.2 Pro model page, GPT-5.2 Pro costs $21 per 1 million input tokens そして $168 per 1 million output tokens. OpenAI does not list a cached-input discount for this model. Tool calls can add separate charges when a workflow uses paid tools.
Those are usage-based API prices, not a monthly ChatGPT plan. A long output can cost much more than the input, so set output limits and test representative workloads before keeping GPT-5.2 Pro in a production route.
How to use GPT-5.2 Pro through the API
- Open the official GPT-5.2 Pro model page and select Try in Playground, or use an OpenAI API project with billing enabled.
- Send the request through the Responses API with
model="gpt-5.2-pro". - 選ぶ
中,高, あるいはxhighreasoning effort based on measured task quality, latency, and cost. - Use background mode for requests that may take several minutes, as OpenAI recommends for this slower model.
- Log token usage, latency, errors, and task success before deciding whether the older model is worth retaining.
from openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-5.2-pro",
reasoning={"effort": "high"},
input="Review this risk model and list any unsupported assumptions."
)
print(response.output_text)This example shows the model identifier and reasoning setting, but it does not make a high-stakes result reliable by itself. Financial, legal, medical, scientific, and security-sensitive conclusions still require domain review and source verification.
What GPT-5.2 Pro does well—and where it falls short
Reasonable use cases
- Reproducing or regression-testing an application already validated on the December 2025 snapshot.
- Reviewing long text-and-image inputs when a 400,000-token context window is sufficient.
- Running difficult reasoning tasks where your own evaluations show a measurable advantage over cheaper models.
- Maintaining a temporary compatibility route while migrating prompts, tools, and quality tests to a successor.
Important limitations
- It is not current in ChatGPT. No ChatGPT plan selection can bring it back.
- It is a previous-generation API model. Newer OpenAI families have superseded its position.
- It can be slow. OpenAI warns that difficult requests may take several minutes.
- Its knowledge is not live. The published cutoff is August 31, 2025.
- Structured outputs are not supported. Choose a newer compatible model if schema-constrained output is essential.
- Cost is high. The output-token price is $168 per million tokens.
- Large context is not perfect recall. A 400,000-token window does not guarantee that every detail will be used correctly.
- It is not an autonomous professional. “Pro” describes a model tier, not legal, financial, medical, or engineering accountability.
GPT-5.2 Pro vs its successors
OpenAI’s naming changed after GPT-5.5. The GPT-5.2 Pro model page recommends GPT-5.5プロ as the newer dedicated Pro model. The current GPT-5.6 family then moved Pro behavior into an execution setting: use gpt-5.6-sol, あるいは gpt-5.6 alias that routes to it, with reasoning.mode: "pro" for the current frontier route described in OpenAI’s guide.
| Model route | Official position | Context / max output | Published input / output rate | Best reason to choose it |
|---|---|---|---|---|
gpt-5.2-プロ | Previous Pro model | 400,000 / 128,000 tokens | $21 / $168 per 1M tokens | Compatibility with an existing evaluated workflow |
gpt-5.5-pro | Direct dedicated-Pro successor | 1,050,000 / 128,000 tokens | $30 / $180 per 1M tokens | A newer fixed Pro model route when that exact model is required |
gpt-5.6-sol with Pro mode | Current frontier route for the hardest API work | 1,050,000 / 128,000 tokens | $5 / $30 per 1M tokens; Pro mode can consume more tokens | New quality-first work that can tolerate extra latency and token usage |
For cost and plan details on the newest family, see the GPT-5.6 価格ガイド. For a broader transition view, compare GPT-5.6 and GPT-5.5 in practical workflows, or review GPT-5.5 access, pricing, and limitations.
Troubleshooting GPT-5.2 Pro access
| 問題点 | Likely explanation | どうすればいい |
|---|---|---|
| The model is missing in ChatGPT | This is expected after the June 12, 2026 retirement. | Use a current ChatGPT model; a higher ChatGPT plan will not restore GPT-5.2 Pro. |
| The API rejects the model | Your API project may lack billing, the required usage tier, or account-level access. | Check the API project, billing state, model page, and returned error. Do not treat ChatGPT subscription status as API access. |
| The request times out | GPT-5.2 Pro can take several minutes on difficult prompts. | Use background mode and poll for completion instead of relying on a short synchronous timeout. |
| JSON does not match the requested schema | The model does not support Structured Outputs. | Validate and repair output in your application, or move the workflow to a current model that supports Structured Outputs. |
| The answer contains stale facts | The published knowledge cutoff is August 31, 2025. | Provide current first-party sources or connect an appropriate retrieval or search tool, then verify the final claim. |
Should you still use GPT-5.2 Pro?
Use GPT-5.2 Pro only when you have a concrete compatibility reason and evaluations showing that migration would reduce quality or break behavior. Its official API listing means it remains usable; it does not mean it is the best default for a new application.
For a new build, start with GPT-5.6 and test standard mode before Pro mode. Add Pro mode only when the quality gain matters enough to justify extra latency and token usage. For an existing GPT-5.2 Pro route, compare the same production tasks across GPT-5.2 Pro, GPT-5.5 Pro, and GPT-5.6, then migrate on measured task success rather than model-name prestige.
よくある質問
Is GPT-5.2 Pro still available in 2026?
Yes in the OpenAI API, but no in ChatGPT. As of July 23, 2026, OpenAI still lists gpt-5.2-プロ for the Responses API. OpenAI removed GPT-5.2 Pro from ChatGPT on June 12, 2026.
Is GPT-5.2 Pro still OpenAI’s most powerful model?
No. OpenAI calls GPT-5.2 Pro a previous Pro model. GPT-5.5 Pro is its direct dedicated-Pro successor, and GPT-5.6 is the current flagship family for new API workloads.
Which ChatGPT plan includes GPT-5.2 Pro?
None. GPT-5.2 Pro is no longer available in ChatGPT on any current plan. ChatGPT subscriptions and API billing are separate, so buying ChatGPT Pro does not unlock the older API model.
How do I access GPT-5.2 Pro now?
Use the official OpenAI model page’s Playground route or send a Responses API request with model="gpt-5.2-pro". Your API project needs eligible access, billing, and sufficient rate limits.
What is the GPT-5.2 Pro API price?
As of July 23, 2026, OpenAI lists $21 per 1 million input tokens and $168 per 1 million output tokens. This is usage-based API pricing, not a monthly ChatGPT subscription.
What are GPT-5.2 Pro’s context and output limits?
The official model page lists a 400,000-token context window and a 128,000-token maximum output. These are technical ceilings; actual usable length depends on the request and system limits.
Does GPT-5.2 Pro support images and Structured Outputs?
It accepts text and image input and returns text. OpenAI lists streaming and function calling as supported, but Structured Outputs, fine-tuning, predicted outputs, audio, and video are not supported.
What should replace GPT-5.2 Pro in a new project?
Start with GPT-5.6. For the hardest quality-first work, OpenAI recommends GPT-5.6 Sol in the Responses API with Pro mode. Keep GPT-5.2 Pro only when compatibility testing gives a specific reason.
The bottom line
GPT-5.2 Pro has moved from flagship to compatibility model. It remains an officially listed, expensive, slow, quality-first Responses API option with a 400,000-token context window, but it is gone from ChatGPT and is no longer OpenAI’s leading professional model. Most new work should begin with GPT-5.6; existing GPT-5.2 Pro applications should migrate only after representative evaluations confirm the replacement.
