OpenAI Pro model
GPT-5.5 Pro
A higher-compute version of GPT-5.5 designed to think harder and produce smarter, more precise responses on difficult professional problems.
- Context window
- 1.05M
- tokens
- Max output
- 128K
- tokens
- Input
- $30.00
- per 1M tokens
- Cached input
- No discount
- same token rate
- Output
- $180.00
- per 1M tokens
01 / Overview
What GPT-5.5 Pro Is
GPT-5.5 Pro is a higher-compute version of GPT-5.5 built for difficult problems where smarter and more precise answers are worth additional latency and substantially higher inference cost.
A Pro Model Optimized for Answer Quality
OpenAI describes GPT-5.5 Pro as a version of GPT-5.5 that uses more compute to think harder and provide consistently better answers. That makes its role different from a general-purpose production model.
The Pro tier is most relevant when the difficult part of the workload is reasoning itself. A task may involve ambiguous evidence, several interacting constraints, long technical context, or a decision that requires careful synthesis before the final response is produced.
This extra compute changes how GPT-5.5 Pro should be evaluated. Faster response time and low cost are not its primary advantages. Instead, the useful question is whether it meaningfully improves accuracy, precision, consistency, or the amount of human correction required on the hardest cases in your application.
OpenAI also notes that some GPT-5.5 Pro requests may take several minutes to complete and recommends background mode to avoid timeouts on long-running work.
- Uses more compute than standard GPT-5.5 to tackle tougher problems.
- Focused on smarter, more precise, and more consistent responses.
- Supports a 1.05M-token context window.
- Uses high reasoning effort by default.
- Intended for workloads where answer quality outweighs latency and token cost.
- Provider
- OpenAI
- Family
- GPT-5.5
- Tier
- Pro
- Knowledge cutoff
- Dec 1, 2025
- Input modalities
- Text, Image
- Output modality
- Text
- Default reasoning
- High
02 / Use cases
When GPT-5.5 Pro Makes Sense
GPT-5.5 Pro is best reserved for tasks where a higher-quality answer has enough value to justify slower execution and a much higher cost per token.
Route the Hardest Cases to Pro
A production system rarely needs Pro-level compute for every request. Routine extraction, rewriting, classification, and simple tool calls are usually better candidates for cheaper models.
GPT-5.5 Pro becomes interesting when the workload reaches the point where small reasoning errors create significant downstream cost. This can include difficult technical investigations, complex professional analysis, high-value document synthesis, or decisions based on large and partially conflicting source sets.
The model can also be used as an escalation tier. A lower-cost model handles normal traffic, while cases with higher complexity, uncertainty, or business value are routed to GPT-5.5 Pro.
This approach makes evaluation especially important. A Pro model should demonstrate a measurable improvement on a deliberately difficult test set rather than simply appearing stronger on easy prompts that other models already solve reliably.
- Difficult professional analysis with multiple constraints.
- Complex technical investigations and architecture reasoning.
- High-value document synthesis where precision matters.
- Escalation workflows for cases that fail a lower-cost model.
- Tasks where reducing human correction can justify substantially higher inference cost.
- 01
Hard technical problems
Reason across large technical contexts, competing constraints, and incomplete evidence before producing a recommendation.
- 02
High-value analysis
Use extra reasoning compute when an inaccurate conclusion would create meaningful review or operational cost.
- 03
Complex synthesis
Combine long source material into a precise, internally consistent professional result.
- 04
Escalated requests
Route only the most difficult cases to Pro after a cheaper model or evaluation layer identifies higher complexity.
03 / Pricing
GPT-5.5 Pro Pricing
GPT-5.5 Pro is priced at $30.00 per million input tokens and $180.00 per million output tokens, placing it firmly in a premium, quality-first model tier.
Cost Per Correct Outcome Matters More Than Cost Per Call
At this price level, GPT-5.5 Pro should not be evaluated using token price alone. The relevant question is whether its additional reasoning quality reduces enough retries, review cycles, errors, or human intervention to justify the premium.
A 10,000-token input with a 2,000-token output costs approximately $0.66 before any tool-specific charges. Larger documents and long generated reports can increase that quickly.
Unlike many other OpenAI models, GPT-5.5 Pro does not offer a cached-input discount. Repeated prompt prefixes therefore do not receive a lower cached-token rate on the published model pricing.
Regional processing endpoints also carry a 10% uplift for GPT-5.5 Pro where regional processing is used.
- $30.00 per 1M input tokens.
- $180.00 per 1M output tokens.
- No cached-input discount.
- Regional processing endpoints carry a 10% pricing uplift.
- Tool-specific calls may introduce additional fees.
1M tokens · USD
- Input
- $30.00
- Cached input
- No discount
- Output
- $180.00
Example: 10K input + 2K output
- Input cost
- $0.3000
- Output cost
- $0.3600
- Estimated total
- $0.6600
04 / Context
A 1.05M-Token Window for Difficult, Context-Heavy Problems
GPT-5.5 Pro provides a 1,050,000-token context window and supports up to 128,000 output tokens, allowing difficult reasoning tasks to incorporate substantial source material in a single working context.
Large Context Is Valuable When Relationships Matter
The main benefit of a large context window is not simply the ability to send more text. It is the ability to keep related evidence together when the model needs to reason across it.
For a technical investigation, the context may include code, logs, architecture documents, issue history, and previous attempts. For a professional analysis, it may contain several reports, source documents, assumptions, and constraints that must remain consistent with one another.
GPT-5.5 Pro is especially relevant when reducing the context would remove relationships that are important to the reasoning process. Its large context gives the model room to work across those relationships before producing the final answer.
However, there is no cached-input discount for this model. Repeatedly sending very large source sets can therefore become expensive, and context selection remains an important cost-control mechanism even when the full window is technically available.
- 1,050,000-token context window.
- Maximum output of 128,000 tokens.
- Useful for difficult tasks that depend on many related sources.
- Large contexts can materially increase cost because cached input receives no discount.
- Context relevance should be evaluated alongside context capacity.
Context window
1,050,000
Max output
128,000
A Pro workload may include technical documents, code, retrieved evidence, policies, prior analysis, and other source material required for careful reasoning.
05 / Reasoning
High-Effort Reasoning by Design
GPT-5.5 Pro supports medium, high, and xhigh reasoning effort, with high as the default. Lower settings such as none and low are not available.
Pro Starts Where General-Purpose Models Scale Up
The available reasoning range reflects the model's purpose. GPT-5.5 Pro is not designed for zero-reasoning or lightweight inference. Even its lowest supported setting, medium, assumes that the task benefits from deliberate reasoning.
The default high setting is a practical signal of how OpenAI expects the model to be used: difficult problems that warrant significant compute. xhigh can be evaluated for the subset of tasks where even more reasoning materially improves the final result.
This makes GPT-5.5 Pro a poor candidate for workflows that require extremely low latency or very inexpensive routine processing. It is better treated as a specialized tier for cases where deeper reasoning is itself the feature.
OpenAI notes that some requests may take several minutes, so long-running calls should be designed with background mode rather than assuming a conventional synchronous request will always complete quickly.
Difficult professional task
Medium can establish a lower-compute Pro baseline for tasks that still require substantial reasoning.
Maximum precision case
Test high or xhigh when the problem is difficult enough that additional reasoning quality can justify more latency and compute.
06 / Capabilities
Tools for Deep, Grounded Workflows
GPT-5.5 Pro supports function calling, structured outputs, text and image input, and a focused set of Responses API tools for research, retrieval, code execution, and external integrations.
A Different Tool Profile from Standard GPT-5.5
The Pro model supports web search and file search, which makes it useful for workflows that need to ground reasoning in external or retrieved information.
It also supports image generation, code interpreter, hosted shell, and MCP. These capabilities let an application combine high-effort reasoning with executable analysis, command-line workflows, and external tool integrations.
However, GPT-5.5 Pro does not support every tool available on the standard GPT-5.5 model. OpenAI currently lists Apply Patch, Skills, computer use, and Tool Search as unsupported for GPT-5.5 Pro.
Streaming is also not supported. Applications should therefore plan around complete responses rather than progressively rendering model output, and long-running tasks may need background mode.
- Text and image input with text output.
- Function calling.
- Structured outputs.
- Web search and file search.
- Image generation.
- Code interpreter and hosted shell.
- MCP support.
- Streaming is not supported.
- Apply Patch, Skills, computer use, and Tool Search are not supported.
- Fine-tuning is not supported.
- Supported
Function calling
Connect high-effort reasoning to application-defined functions.
- Supported
Structured outputs
Return predictable machine-readable results for downstream systems.
- Supported
Search and retrieval
Use web search and file search to ground difficult reasoning tasks in external evidence.
- Supported
Code execution
Use code interpreter and hosted shell for executable analysis and technical workflows.
- Supported
MCP
Connect the model to supported external tools and data through MCP.
- Not listed
Streaming
Progressive token streaming is not supported for GPT-5.5 Pro.
07 / Evaluation
Strengths and Limitations
GPT-5.5 Pro is a specialized precision-first model: its value comes from improving difficult answers, not from being the fastest or most economical option for general production traffic.
Strengths
Higher-compute reasoning
OpenAI designed GPT-5.5 Pro to think harder and provide smarter, more precise responses than standard GPT-5.5.
Large context capacity
A 1.05M-token window provides substantial room for complex evidence, documents, code, and professional source material.
Precision-first configuration
Reasoning starts at medium and defaults to high, aligning the model directly with difficult tasks rather than lightweight inference.
Grounded and executable workflows
Web search, file search, code interpreter, hosted shell, function calling, and MCP support substantial research and technical tasks.
What to consider
Very high token cost
At $30 per 1M input tokens and $180 per 1M output tokens, GPT-5.5 Pro should be reserved for workloads where improved precision creates clear value.
No cached-input discount
Repeated prompt context is not billed at a reduced cached-token rate.
Potentially long latency
OpenAI states that difficult requests can take several minutes and recommends background mode to avoid timeouts.
Reduced tool surface
Apply Patch, Skills, computer use, and Tool Search are not supported, and streaming is unavailable.
Test precision where it matters
Evaluate GPT-5.5 Pro on your hardest tasks
Run the difficult cases that justify a Pro model and inspect response quality, token usage, request cost, context consumption, and reasoning settings before using it in production.
Start FreeConnect your own OpenAI API key and evaluate GPT-5.5 Pro against the exact prompts and source material where additional reasoning quality could create measurable value.
Common Questions
What is GPT-5.5 Pro?
GPT-5.5 Pro is a higher-compute version of GPT-5.5 designed to think harder and produce smarter, more precise responses on difficult problems.
How much does GPT-5.5 Pro cost?
GPT-5.5 Pro costs $30.00 per 1M input tokens and $180.00 per 1M output tokens. OpenAI does not offer a cached-input discount for this model.
What is the context window of GPT-5.5 Pro?
GPT-5.5 Pro has a 1,050,000-token context window and supports up to 128,000 output tokens.
What reasoning levels does GPT-5.5 Pro support?
GPT-5.5 Pro supports medium, high, and xhigh reasoning effort. High is the default.
What is the knowledge cutoff for GPT-5.5 Pro?
OpenAI lists December 1, 2025 as the knowledge cutoff for GPT-5.5 Pro.
Does GPT-5.5 Pro support image input?
Yes. GPT-5.5 Pro accepts text and image input and produces text output. Direct audio and video modalities are not supported.
Does GPT-5.5 Pro support streaming?
No. OpenAI currently lists streaming as unsupported for GPT-5.5 Pro.
What tools does GPT-5.5 Pro support?
OpenAI currently lists web search, file search, image generation, code interpreter, hosted shell, and MCP as supported Responses API tools. Apply Patch, Skills, computer use, and Tool Search are not supported.
Why can GPT-5.5 Pro requests take longer?
GPT-5.5 Pro uses more compute to reason through difficult problems. OpenAI notes that some requests can take several minutes and recommends background mode to avoid request timeouts.
What workloads are a good fit for GPT-5.5 Pro?
GPT-5.5 Pro is best suited to difficult professional analysis, complex technical reasoning, high-value document synthesis, and escalation workflows where improved precision can justify premium cost and longer latency.
Can GPT-5.5 Pro be fine-tuned?
No. The current OpenAI model documentation lists fine-tuning as unsupported for GPT-5.5 Pro.
Model information
Last updated
The specifications and prices on this page are based on the official OpenAI documentation for GPT-5.5 Pro. Provider pricing, endpoints, supported tools, rate limits, regional processing, and availability may change, so production assumptions should be checked against the latest provider documentation.