Enterprise+, Proposal Pro, QPilot Essential
QorusDocs QPilot AI Assistant leverages powerful language models to deliver accurate, intelligent, and context-aware responses to your content queries.
This guide outlines the AI models currently supported by QPilot, their capabilities, and how they differ, helping you better understand the technology behind QPilot’s AI features, including Ask (Smart Search), QPilot Chat, AI Text Smart Fields, and Smart Skills.
In this article, we’ll cover:
- How to set the AI Model
- Choosing the best AI model and when to use them
- Supported AI Models
- Model Update Policy
How to set the AI Model
Admins can choose which AI model QPilot uses at the hub level, and can override that choice for an individual Smart Skill. If you choose a different model for a Smart Skill compared to the one used in Hub Management Settings, then the model set at the Smart Skill level will be used when users execute that Smart Skill.
Set the AI Model your hub uses
You can select any supported option, including the new Default, from the AI Model dropdown in Hub Settings:
Override the hub AI model selected for a Smart Skill
You can override the AI model you selected for your hub for each Smart Skill you configure. To do this:
- For a new Smart Skill, you will find the AI model dropdown menu in the configuration screen.
-
For an existing Smart Skill, go to Manage > Smart Skills, then click the gear icon next to the Smart Skill you want to change the AI model for and click Edit.
Scroll down to find the AI model selection option:
Note that the temperature parameter is not configurable for the GPT-5 series models (GPT-5, GPT-5-mini, GPT-5.1, GPT-5.2, GPT-5.4-mini, and GPT-5.4). These models use a fixed temperature to ensure consistent output quality and maintain their optimized balance between determinism and creativity. You can learn more about AI temperature in Smart Skills Configuration.
Choosing the best AI model and when to use them
For most teams, we recommend Default. It uses the model QorusDocs has selected and validated for the work QPilot supports, and we keep it current for you, so you do not have to track model releases and retirements. Default currently uses GPT-5.2.
If you would rather choose a specific model, here is how the options compare for the proposal, RFP, and content work QPilot is built for:
- GPT-5.2, the balanced choice. The strongest all-round balance of reasoning quality and speed for QPilot’s use cases, and the model we selected as Default after evaluation. Start here unless you have a specific reason not to.
- GPT-5.4, maximum capability. Best for the most complex, high-stakes work where depth of reasoning matters more than speed, such as detailed compliance or legal analysis.
- GPT-5.4-mini, speed and scale. Best for high-volume, latency-sensitive work where fast, consistent output matters most.
- Earlier GPT-5 models (GPT-5, GPT-5.1, GPT-5-mini). Still available and a reasonable choice if your team has standardised on one, though GPT-5.2 is the better balance for most use cases.
- GPT-4 models (GPT-4o, GPT-4.1). Earlier-generation models that remain available for existing configurations. We recommend moving to Default or a GPT-5 series model.
Supported AI Models
QPilot offers the following options in the AI Model dropdown, including the new QorusDocs–managed Default:
- GPT-5.2: the balanced, recommended choice
- GPT-5.4: maximum capability for the most demanding work
- GPT-5.4-mini: fast and efficient for high-volume work
- GPT-5.1
- GPT-5
- GPT-5-mini
These models offer varying levels of performance, context understanding, and processing power.
Each model also has a maximum session capacity, measured in tokens, that determines how much a single QPilot Chat session can hold. To understand how that capacity is used and how to stay within it, see QPilot Chat Capacity Indicator.
Default (QorusDocs-managed)
Default is the recommended model. When you select Default, QPilot uses a current-generation model that QorusDocs manages and keeps up to date on your behalf, rather than a specific model you select and maintain yourself. The same Default applies consistently across your Hub and Smart Skills.
Choosing Default means you no longer need to track the continual churn of AI models, where providers regularly release new versions and retire older ones. You stay on a current-generation model automatically.
Default currently uses GPT-5.2. We will keep it current as part of our normal product releases, and we will update this article whenever the model behind Default changes.
Session capacity: matches the model currently behind Default (GPT-5.2), see GPT-5.2 below.
GPT-5 series
GPT-5.2: the balanced, recommended choice
GPT-5.2 is the balanced, recommended choice in the GPT-5 series, and the model QorusDocs selected as Default after evaluation across the proposal, RFP, and content use cases QPilot supports. It offers the strongest all-round balance of reasoning quality and speed for that work. This is the model currently behind Default.
Best for:
- A strong choice for most work, start here unless you have a specific reason to choose another model
- Everyday professional work that benefits from current-generation reasoning
- Proposal drafting, summarisation, and detailed content queries
Context window limit: 400,000 tokens. Max output window: 128,000 tokens.
GPT-5.4: maximum capability for the most demanding work
GPT-5.4 is the latest frontier model, bringing together advances in reasoning and complex analysis for the most demanding, high-stakes work, where depth of reasoning matters more than speed.
Best for:
- The most complex, high-stakes tasks, such as detailed compliance or legal analysis
- Deep reasoning, detailed research, and multi-step problem-solving
Context window limit: 1,050,000 tokens. Max output window: 128,000 tokens.
GPT-5.4-mini: fast and efficient for high-volume work
GPT-5.4-mini is a faster, lower-cost member of the GPT-5.4 family, built for high-volume, latency-sensitive work where fast, consistent output matters most while retaining strong everyday quality.
Best for:
- High-volume, scaled interactions
- Fast, reliable everyday document support and chat
Context window limit: 400,000 tokens. Max output window: 128,000 tokens.
GPT-5.1
GPT-5.1 is an earlier GPT-5 series model that remains available. It is a reasonable choice if your team has standardised on it, though GPT-5.2 offers the better balance for most QPilot use cases.
Context window limit: 400,000 tokens. Max output window: 128,000 tokens.
GPT-5
GPT-5 is an earlier GPT-5 series model that remains available, well suited to high-stakes, complex work. For new configurations we recommend GPT-5.2 or GPT-5.4.
Context window limit: 400,000 tokens. Max output window: 128,000 tokens.
GPT-5-mini
GPT-5-mini is an earlier, lighter GPT-5 series model that remains available, designed for faster, lower-cost responses. For new work, GPT-5.4-mini offers a current-generation alternative for speed and scale.
Context window limit: 400,000 tokens. Max output window: 128,000 tokens.
GPT-4 series
GPT-4.1
GPT-4.1 is an earlier-generation OpenAI model that remains available for existing configurations. For new work, we recommend Default or a GPT-5 series model.
Context window limit: 1,047,576 tokens. Max output window: 32,768 tokens.
GPT-4o
GPT-4o is an earlier-generation OpenAI model that remains available for existing configurations. For new work, we recommend Default or a GPT-5 series model.
Context window limit: 128,000 tokens. Max output window: 16,384 tokens.
Model Update Policy
The AI landscape is rapidly evolving. QorusDocs ensures that QPilot remains up to date with the latest advancements to deliver the most relevant, accurate, and efficient AI-powered assistance.
When you use Default, QorusDocs manages the underlying model for you and moves it forward to a current-generation model over time. We validate a new model before it becomes the Default, and we announce any change to the model behind Default, along with other model updates, via our release notes. This article always reflects the model currently behind Default.
Updates will be communicated via release notes.
Comments
Please sign in to leave a comment.