> ## Documentation Index
> Fetch the complete documentation index at: https://developer.box.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Supported AI models

> Core and customer-enabled AI models available in Box AI, with access levels, capability tiers, and API names.

Box supports a variety of AI models, categorized along two dimensions: [access level](#access-levels) and [capability tier](#capability-tiers).

## Access levels

| Level | Description | Configuration Required |
| - | - | - |
| [**Core models**](#core-models) | Built into Box AI and available by default for all customers | None |
| [**Customer-enabled models**](#customer-enabled-models) | Require activation by Box admins in the Admin Console or a request to Box. Some models may be subject to additional terms or pricing. | Yes |

## Capability tiers

| Tier | Description | Best For |
| - | - | - |
| **Standard models** | High-speed, cost-efficient performance | Basic summarization, Q\&A, structured data extraction from shorter or simpler documents. High-volume, low-complexity use cases |
| **Premium models** | Advanced reasoning, larger context windows, better performance on complex content | Multi-step reasoning, large taxonomies, lengthy or unstructured documents, domain-specific content |
| **Ultra models** | Highest-tier intelligence for complex, long-running, multi-step content work | Orchestrating long-running, multi-stage workflows, multiple knowledge bases, and deep domain-specific reasoning |

> [!NOTE]
>
> A model can be both customer-enabled and premium, or core and standard. In other words, access level and capability tiers are independent categorizations (for example, models can be either capability tier regardless of access level). The two categorizations are complementary.

## Using models

How to use the supported AI models:

* Get the default AI agent configuration.
* Override the AI agent configuration used in `POST 2.0/ai/ask`, `POST 2.0/ai/text_gen`, `POST 2.0/ai/extract`, `POST 2.0/ai/extract_structured` endpoints.

When using the `model` parameter in your API calls, use the **API Name** visible on each model card listed below.

For example, to get the AI agent configuration for a specific model, use the model parameter and provide the `openai__gpt_5_mini` API name. Make sure you use **two underscores** after the provider name.

> [!NOTE]
>
> The list may change depending on the model availability.

A free developer account gives you access to these AI models through the Box AI API. Start making API calls to test summarization, extraction, and more.

[https://account.box.com/signup/developer](https://account.box.com/signup/developer)
[https://account.box.com/developers/console](https://account.box.com/developers/console)

## Core models

Box AI is powered by the following AI models. These models are integrated with Box AI to facilitate various use cases while adhering to enterprise grade standards. Below, you’ll find information about each model, including its capabilities, intended applications, and applicable usage guidelines.

> [!WARNING]
>
> Models offered in **Preview** mode have not been fully performance-tested at scale. You may experience variability in output quality, availability, and accuracy.

> [accordion] OpenAI
>
> > [card] openai\_\_gpt\_6\_1\_sol
> >
> > Premium model in the GPT-6 family with near-Astra performance at a lower cost
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_6\_1\_terra
> >
> > Standard model for visual reasoning, computer use, and writing
> >
> > StandardCompatible with Box Agent
>
> > [card] openai\_\_gpt\_6\_sol
> >
> > Premium model in the GPT-6 family, built for complex, high-stakes professional work
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_6\_luna
> >
> > Fast, cost-efficient model in the GPT-6 family for high-volume, general-purpose work
> >
> > StandardCompatible with Box Agent
>
> > [card] openai\_\_gpt\_6\_astra
> >
> > Ultra model from OpenAI
> >
> > UltraCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_6\_sol
> >
> > Most capable GPT-5.6 model, built for complex, high-stakes professional work
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_6\_terra
> >
> > Multimodal model for broad general-purpose work and most coding tasks
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_6\_luna
> >
> > Cost-efficient GPT-5.6 model for high-volume workloads
> >
> > Standard
>
> > [card] openai\_\_gpt\_5\_5
> >
> > Advanced model for the most complex professional work
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_4
> >
> > Multimodal model for both broad general-purpose work and most coding tasks
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_4\_mini
> >
> > Faster, more cost-efficient version of GPT-5.4
> >
> > Standard
>
> > [card] openai\_\_gpt\_5\_2
> >
> > Multimodal model for coding and agentic tasks across various industries
> >
> > PremiumCompatible with Box Agent
>
> > [card] openai\_\_gpt\_5\_1
> >
> > Multimodal model with enterprise-grade performance and adaptive reasoning
> >
> > Premium
>
> > [card] openai\_\_gpt\_5
> >
> > Multimodal model with advanced reasoning and long-context understanding
> >
> > Premium
>
> > [card] openai\_\_gpt\_5\_mini
> >
> > Model designed for well-defined tasks and precise prompts, suitable for lightweight tasks
> >
> > Standard
>
> > [card] azure\_\_openai\_\_text\_embedding\_ada\_002
> >
> > 2nd-generation embedding model for text search, code search, and sentence similarity
> >
> > EmbeddingsStandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2

> [accordion] Google
>
> > [card] google\_\_gemini\_4\_hydrogen
> >
> > Fast model for everyday coding and agentic workflows
> >
> > PremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_4\_carbon
> >
> > Premium model for long-horizon agentic workflows across knowledge work, engineering, and scientific reasoning
> >
> > PremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_4\_argon
> >
> > Multimodal reasoning model for software engineering, knowledge work, and cybersecurity
> >
> > PremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_8\_flash
> >
> > Fast, multimodal model for software engineering and agentic knowledge workflows
> >
> > StandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_7\_flash
> >
> > Fast, multimodal model with improved reasoning and agentic video understanding
> >
> > StandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_5\_flash
> >
> > Fast, multimodal model with advanced reasoning for demanding enterprise tasks
> >
> > PremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_5\_flash\_lite
> >
> > Multimodal model for high-speed, cost-efficient tasks
> >
> > StandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_1\_flash\_lite
> >
> > Gemini multimodal model for high-speed and cost-efficient tasks
> >
> > StandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4
>
> > [card] google\_\_gemini\_2\_5\_pro
> >
> > Multimodal model with a 1 million token context window and advanced reasoning capabilities
> >
> > PremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5
>
> > [card] google\_\_gemini\_2\_5\_flash
> >
> > Multimodal model offering well-rounded capabilities, including thinking capabilities
> >
> > StandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4DoD IL5

> [accordion] Anthropic
>
> > [card] aws\_\_claude\_haiku\_5\_5
> >
> > Fastest, most efficient Claude 5.5 model for high-volume, latency-sensitive work
> >
> > StandardISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_sonnet\_5\_5
> >
> > Smarter, more efficient Sonnet model for coding and knowledge work
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_opus\_5\_5
> >
> > Premium model for coding, knowledge work, and long-running tasks
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_opus\_5
> >
> > Premium model built for complex reasoning and long-horizon agentic work
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_sonnet\_5
> >
> > Powerful, versatile model built for daily use, scaled production, and complex tasks
> >
> > PremiumISMAPFedRAMP ModerateCompatible with Box Agent
>
> > [card] aws\_\_claude\_4\_8\_opus
> >
> > Premium model from Anthropic, hosted on Amazon Web Services
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_7\_opus
> >
> > Multimodal model for coding, enterprise agents, and professional work
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_6\_opus
> >
> > Multimodal model for complex tasks with a 1 million token context window
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_6\_sonnet
> >
> > Multimodal model for complex tasks with a 1 million token context window
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_5\_opus
> >
> > Premium model combining maximum intelligence with practical performance
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_5\_sonnet
> >
> > Model that excels at complex agents, coding, and autonomous multi-step workflows
> >
> > PremiumISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_4\_5\_haiku
> >
> > Fast and cost-efficient model with strong reasoning, best for low-latency applications
> >
> > StandardISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent

> [accordion] Meta
>
> > [card] ibm\_\_llama\_4\_maverick
> >
> > Multimodal model with a 1 million token context window and advanced reasoning capabilities
> >
> > Standard

> [accordion] Mistral AI
>
> > [card] ibm\_\_mistral\_medium\_2505
> >
> > High-performance enterprise model for coding and advanced reasoning
> >
> > PreviewStandard
>
> > [card] ibm\_\_mistral\_small\_3\_1\_24b\_instruct\_2503
> >
> > Fast open-source multimodal model with low latency
> >
> > PreviewStandard

> [accordion] NVIDIA
>
> > [card] azure\_\_nemotron\_3\_ultra\_nvfp4
> >
> > Open, text-only reasoning model for complex agentic workflows and long-context analysis
> >
> > StandardCompatible with Box Agent

## Customer-enabled models

Certain Box AI customers may enable additional AI models upon their request or upon the models otherwise being made available through their Admin Console. Use of these models may be subject to additional terms. By selecting a customer-enabled model, the customer acknowledges that their data may be processed by additional [subprocessors][subprocessors] of their choice.

> [!WARNING]
>
> Models offered in **Beta** mode are provided on an as-is basis.

> [accordion] OpenAI
>
> > [card] openai\_\_gpt\_o3
> >
> > Multimodal model, highly efficient in handling complex, multi-step tasks
> >
> > BetaPremium

> [accordion] Google
>
> > [card] google\_\_gemini\_3\_6\_flash
> >
> > Natively multimodal reasoning model in the Gemini 3 Flash series
> >
> > BetaPremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4Compatible with Box Agent
>
> > [card] google\_\_gemini\_3\_1\_pro
> >
> > Model for complex tasks with advanced reasoning and a large context window
> >
> > BetaPremiumISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4
>
> > [card] google\_\_gemini\_3\_flash
> >
> > Multimodal model designed for speed and efficiency across a wide range of tasks
> >
> > BetaStandardISMAPFedRAMP ModerateFedRAMP HighDoD IL2DoD IL4

> [accordion] Anthropic
>
> > [card] aws\_\_claude\_fable\_5\_2
> >
> > Ultra model for demanding reasoning and long-horizon agentic work
> >
> > BetaUltraCompatible with Box Agent
>
> > [card] aws\_\_claude\_fable\_5\_1
> >
> > Ultra model from Anthropic, hosted on Amazon Web Services
> >
> > BetaUltraISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent
>
> > [card] aws\_\_claude\_fable\_5
> >
> > Ultra model from Anthropic, hosted on Amazon Web Services
> >
> > BetaUltraISMAPFedRAMP ModerateDoD IL2Compatible with Box Agent

> [accordion] Meta
>
> > [card] meta\_\_muse\_ember
> >
> > Small multimodal model for high-volume agentic work
> >
> > BetaStandardCompatible with Box Agent
>
> > [card] meta\_\_muse\_spark\_1\_3
> >
> > Multimodal model for general-purpose text and image understanding
> >
> > BetaStandardCompatible with Box Agent
>
> > [card] meta\_\_muse\_spark\_1\_2
> >
> > Multimodal model for general-purpose text and image understanding
> >
> > BetaStandardCompatible with Box Agent

> [accordion] xAI
>
> > [card] xai\_\_grok\_4\_7
> >
> > Multimodal reasoning model for coding, agentic tasks, and knowledge work
> >
> > BetaPremiumCompatible with Box Agent
>
> > [card] xai\_\_grok\_4\_6
> >
> > Multimodal reasoning model for coding, agentic tasks, and knowledge work
> >
> > BetaPremiumCompatible with Box Agent

## Legacy agents

Legacy agents represent the initial generation of Box AI capabilities, designed to handle basic, single-step tasks such as simple Q\&A and document summarization. The table below specifies available models for a particular use case.

| Use case | Provider | Model for Standard Version | Model for Advanced or Enhanced Version |
| - | - | - | - |
| Q\&A for Documents, Notes, and Hubs | OpenAI (Default) | GPT-5 mini | GPT-5.2 |
| | Claude | Claude 4.5 Sonnet | Claude 4.5 Opus |
| | Llama | Llama 4 Maverick | Llama 4 Maverick |

## Box agents

Box agents are the next generation of intelligent assistants built with advanced reasoning and orchestration, capable of planning and executing complex, multi-step workflows across your entire Box content ecosystem. The table below specifies available models for each specialized agent type.

| Agent | Provider | Model for Standard Version | Model for Enhanced Version |
| - | - | - | - |
| Extract Agent | Google | Gemini 3.1 Flash Lite | Gemini 3.7 Flash |
| Security Classification Agent | OpenAI | GPT-6 Luna | N/A |
| | Google | Gemini 3.1 Flash Lite | N/A |
| Threat Analysis Agent | Google | Gemini 3.5 Flash Lite | N/A |

## Retired models

The following models are retired and no longer available in Box AI.

| Model name | API name |
| - | - |
| Azure OpenAI GPT-4.1 | `azure__openai__gpt_4_1` |
| Azure OpenAI GPT-4.1 mini | `azure__openai__gpt_4_1_mini` |
| Azure OpenAI GPT-4o | `azure__openai__gpt_4o` |
| Azure OpenAI GPT-4o mini | `azure__openai__gpt_4o_mini` |
| Claude 4 Opus | `aws__claude_4_opus` |
| Claude 4 Sonnet | `aws__claude_4_sonnet` |
| Titan Text Lite | `aws__titan_text_lite` |
| Gemini 2.0 Flash | `google__gemini_2_0_flash_001` |
| Gemini 2.0 Flash Lite | `google__gemini_2_0_flash_lite_preview` |
| Gemini 3 Pro | `google__gemini_3_pro` |
| Llama 4 Scout | `ibm__llama_4_scout` |
| Llama 3.2 90B Vision | `ibm__llama_3_2_90b_vision_instruct` |
| Grok 3 Beta | `xai__grok_3_beta` |
| Grok 3 mini Beta | `xai__grok_3_mini_beta` |

[azure-ai-mini-4o-model]: https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models?tabs=python-secure#gpt-4o-and-gpt-4-turbo

[vertex-ai-model]: https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models#models

[vertex-ai-gemini-models]: https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models#gemini-models

[vertex-text-models]: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text

[azure-ai-embeddings]: https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models#embeddings

[aws-claude]: https://aws.amazon.com/bedrock/claude/

[aws-titan]: https://aws.amazon.com/bedrock/titan/

[subprocessors]: https://www.box.com/legal/subprocessors
