Models
The AI models and tools that may power your creative work in awen.
Model usage
Awen uses a combination of proprietary, third-party, and ancillary artificial intelligence models to provide its agentic features, including text generation, image generation, video generation, 3D generation, image and video enhancement, document processing, tagging, and related automated workflows.
This glossary explains which models may be used by Awen, how model selection is handled, and how this list is maintained.
User acceptance
By using Awen, users acknowledge and accept that the Awen agent may rely on one or more of the models listed in this glossary.
The model used may depend on the task requested, the technical requirements of the workflow, model availability, performance, cost, safety, latency, and the configuration of Awen's infrastructure at the relevant time.
Updates to the model list
This list is kept up to date as soon as changes are made to Awen's infrastructure, including the addition, replacement, upgrade, removal, or deprecation of any model or model provider.
The specific model used for a given task may vary and may change without requiring a separate notice to users, provided that such use remains consistent with Awen's applicable terms, privacy commitments, and customer agreements.
Text models
Awen may use the following models for text generation, reasoning, conversation, content editing, summarization, translation, analysis, and other text-based tasks:
- GPT 5.6 Terra by OpenAI
- GPT 5.6 Sol Low by OpenAI
- Claude Fable 5 by Anthropic
- GPT 5.6 Sol XHigh by OpenAI
- Gemini 3.1 Pro by Google
- Claude Opus 4.8 by Anthropic
- Claude Sonnet 5 by Anthropic
Image models
Awen may use the following models and tools for image generation, image editing, image transformation, visual asset creation, and image upscaling:
- Nano Banana 2 Lite, Nano Banana 2, and Nano Banana 2 Pro by Google
- GPT Image 2 by OpenAI
- Seedream 5 Pro by ByteDance
- MAI image 2.5 by Microsoft
- Magnific Image Upscaler
Video models
Awen may use the following models and tools for video generation, video transformation, video enhancement, and video upscaling:
- Seedance 2.0 by ByteDance
- Seedance 2.0 Mini by ByteDance
- Gemini Omni Flash by Google
- Kling v3 Pro
- Kling o3 Pro
- Grok Imagine 1.5 by xAI
- Magnific Video Upscaler
3D models
Awen may use the following models for 3D asset generation, 3D reconstruction, and related spatial or visual generation tasks:
- Rodin v2.5 by Hyper3D
Ancillary models
Awen may also use ancillary models to support the operation of the platform. These models may be used for tasks such as document processing, optical character recognition, tagging, classification, extraction, routing, moderation support, formatting, metadata generation, and other infrastructure-level or workflow-support functions.
Ancillary models may include:
- Mistral OCR
- Gemini 3.1 Flash Lite by Google
- Gemini 3.5 Flash by Google
- Claude Sonnet 4.6 by Anthropic
- Quiver Arrow 1.1 and Quiver Arrow 1.1 Max by Quiver
- Photoroom background removal
Model selection and parametering
Unless expressly specified otherwise by the user, Awen is responsible for selecting, routing, configuring, and parametering the models and technical systems used to process a user request.
This means that Awen may determine which model, tool, provider, workflow, parameters, settings, or sequence of technical operations is most appropriate for a given task. This selection may be based on factors such as quality, speed, availability, model capabilities, infrastructure constraints, safety requirements, cost, latency, and compatibility with the requested output.
Users remain entirely able to direct Awen to select a specific model, provider, workflow, or technical route where such instruction is expressly defined by the user. Awen will follow such direction where technically feasible and compatible with the capabilities, limitations, and intended use of the requested model or route.
In some cases, a user may request a model or technical route that is not compatible with the requested task, output format, input type, safety requirements, or available infrastructure. In such cases, Awen may adapt the technical route, select a different model, modify parameters, use additional supporting tools, or otherwise adjust the workflow as necessary to complete the request or preserve the proper functioning of the service.
Users acknowledge that they may not always be informed of the specific model, parameters, or technical route used for each individual operation, unless such information is expressly made available within the Awen interface or required under an applicable customer agreement.
Ongoing maintenance
Awen may update this glossary from time to time to reflect changes in its infrastructure, service providers, technical architecture, or product capabilities.
Updates may include adding new models, replacing existing models, removing deprecated models, or changing the purpose for which a model is used.
Awen will keep this list reasonably up to date so that users have visibility over the categories of models that may be used when interacting with the platform.