簡単な答え： GPT-5.5 API pricing starts at $5 per 1 million input tokens, $0.50 per 1 million cached input tokens, そして $30 per 1 million output tokens. The model supports a 1,050,000-token context window and up to 128,000 output tokens. The important catch is that prompts above 272,000 input tokens trigger higher rates for the entire session.
GPT-5.5 is built for demanding coding and professional work, but its price is not one flat monthly number. Your actual cost depends on how you access it, how many tokens you send and receive, whether repeated input can be cached, and whether a prompt crosses the long-context pricing threshold.
This guide separates those moving parts. It explains API token rates, long-context pricing, Batch and Flex options, practical cost examples, and the difference between paying for a chat plan and paying for API usage.
If you want to use GPT-5.5 without setting up billing and code for a direct API account, GPT-5.5 対 GlobalGPT offers a simpler model-selection route. It also lets you keep other AI models in the same platform, which is useful when a lower-cost model is enough for part of the job.
目次
GPT-5.5の料金はいくらですか？
OpenAI lists GPT-5.5 text-token pricing per 1 million tokens. The official GPT-5.5 model page shows the following base rates:
GPT-5.5 Base API Rates
Source: OpenAI GPT-5.5 model documentation, checked August 3, 2026. Long prompts and regional processing can change the effective rate.
| トークンの種類 | Base price per 1M tokens | 対象範囲 |
|---|---|---|
| インプット | $5.00 | Your prompt, system instructions, conversation history, documents, tool results, and image inputs after tokenization. |
| キャッシュされた入力 | $0.50 | Eligible repeated prompt content that OpenAI recognizes through prompt caching. |
| 出力 | $30.00 | Visible answer tokens plus model-generated reasoning tokens that are billed as output. |
Output is six times the base input rate, so long answers can dominate the bill. A workflow that sends a large document but asks for a short classification may cost less than a small prompt that produces a very long report.
Do not confuse API prices with a subscription. The figures above are metered API rates. They do not mean GPT-5.5 costs $5 per month, and a ChatGPT subscription does not automatically include API credits.
GPT-5.5 Plans and Access Routes
“GPT-5.5 pricing” can refer to three different buying routes. Choose the route before comparing numbers, because each one measures usage differently.
| アクセスルート | お支払い方法 | 最適 | 主な制限 |
|---|---|---|---|
| OpenAI API | Metered token usage, plus applicable tool charges | Apps, automations, agents, production systems, and developers who need usage control | Requires API setup, billing, and cost monitoring |
| ChatGPTプラン | Subscription with plan-specific access and usage limits | People who want a ready-made chat interface | Availability and limits can depend on the current plan, account, region, and rollout |
| グローバルGPT | Platform plan or credits under GlobalGPT’s current terms | People who want GPT-5.5 and other models in one interface without direct API setup | Credits are a platform measure and should not be treated as OpenAI API dollars |
Standard API processing
Standard processing is the baseline for most API cost estimates. Use the $5 input, $0.50 cached-input, and $30 output rates unless your request qualifies for a different processing tier or crosses the long-context threshold.
バッチAPI
OpenAI’s Batch API is designed for asynchronous work that does not need an immediate answer. The official API pricing page lists discounted token rates for eligible batch jobs. It can make sense for large offline evaluations, document labeling, data enrichment, or nightly processing where turnaround time is flexible.
Flex processing
Flex processing trades response-time consistency and availability for lower pricing. It fits non-production or low-priority jobs that can tolerate slower or occasionally unavailable processing.
Priority and regional processing
Priority processing costs more in exchange for a higher-priority service tier. OpenAI also states that regional processing endpoints carry a 10% uplift for GPT-5.5.
How Tokens Affect Your GPT-5.5 Bill
A token is a piece of text or other model input. It is not the same as a word. English text often averages more than one token per word once punctuation, formatting, code, tables, and special characters are included.
cost = (input tokens / 1,000,000 × input rate) + (output tokens / 1,000,000 × output rate)Cached input is calculated at its separate rate. Tool calls may add their own charges, so token math alone may not equal the final request cost when web search, computer use, or another paid tool is involved.
What counts as input and output?
- Input includes prompts, system instructions, history, documents, tool results, structured data, and image inputs after tokenization.
- Output includes the generated response and billable reasoning tokens.
- Cached input covers eligible repeated prompt content recognized through prompt caching.
GPT-5.5 Context Window and the 272K Pricing Threshold
GPT-5.5 supports a 1,050,000トークン・コンテキスト・ウィンドウ そして 128,000 output tokens. The context window is the total working space available to the request, including the input, relevant history, tool results, and generated output.
When a GPT-5.5 prompt contains more than 272,000 input tokens, the full session is charged at 2× the normal input rate そして 1.5× the normal output rate. At base rates, that is $10 per 1 million input tokens and $45 per 1 million output tokens.
Practical warning: Measure the final assembled request, including retrieved content and history, rather than the file size you started with.
GPT-5.5 Cost Examples
These examples use base API rates and exclude paid tool calls, regional uplifts, and long-context multipliers unless noted.
GPT-5.5 Task Cost Calculator
Choose a common workload, then move one slider. The calculator applies a representative per-task token profile automatically.
Estimate per task: 4,000 input + 2,000 output tokens.
100 tasks × ($0.02 GPT-5.5 input + $0.06 GPT-5.5 output) = $8.00
Workload token profiles are editorial estimates, not fixed OpenAI usage guarantees.
| 例 | インプット | 出力 | 推定コスト |
|---|---|---|---|
| Short professional answer | 10,000 | 2,000 | $0.11 |
| Large document review | 100,000 | 10,000 | $0.80 |
| Repeated 100K input with full cache hit | 100,000 cached | 10,000 | $0.35 |
| Long-context session above 272K | 300,000 | 20,000 | $3.90 |
How to Lower GPT-5.5 Costs
- Route simple work to a cheaper model.
- Keep prompts below 272K when practical.
- Reuse stable prompt prefixes.
- Set an output limit.
- Use Batch for asynchronous jobs.
- Track tokens by task.
- Separate tool costs.
と グローバルGPT, you can keep GPT-5.5 available for demanding work and choose another model when speed or lower usage matters more.
Who Should Use GPT-5.5?
GPT-5.5 makes the most sense when the cost of a weaker answer is higher than the extra token price. Examples include complex code changes, long technical analysis, difficult professional writing, multi-step tool use, and large-document work.
It may be unnecessary for simple classification, basic rewriting, short summaries, or high-volume tasks with predictable rules.
よくある質問
How much is GPT-5.5 per 1 million tokens?
The base API price is $5 per 1 million input tokens, $0.50 per 1 million cached input tokens, and $30 per 1 million output tokens.
Is GPT-5.5 included in ChatGPT Plus?
Model access and usage limits can change by plan, account, region, and rollout. Check the live plan page and your model picker. ChatGPT subscriptions and API billing are separate.
GPT-5.5のコンテキストウィンドウとは何ですか？
OpenAI lists a 1,050,000-token context window and a 128,000-token maximum output.
Why does GPT-5.5 cost more above 272K tokens?
Input is charged at 2× and output at 1.5× for the entire session because processing very large contexts requires more compute.
Does a ChatGPT subscription include API tokens?
No. ChatGPT subscriptions and API usage are billed separately.
Are GlobalGPT credits the same as OpenAI API dollars?
No. GlobalGPT credits follow GlobalGPT’s own plan and usage rules.
