AI models: selection, capabilities, Auto and access
Choose an AI model for writing, research, coding or files. Use the model picker, switch between models, and understand when to use Auto or a named model.
Enlarge preview
Choose a model for the task
Choose or switch a model
Open Chat and select the model name to open the picker. Search for the model you want, select it, then send your message. You can switch models between messages. The library includes Arya and families such as GPT, Claude, Gemini and DeepSeek, alongside media models such as Veo.
Choose capabilities for your task
Match the model to the material you want to use and the result you need. Model details show supported inputs, outputs, tools and limits.
What to look for in an AI model
| Detail | How to use it |
|---|---|
| Writing and editing | Use a text model to draft, summarize or rewrite. Compare whether it keeps your source facts and follows the requested tone and format. |
| Research | For current information, choose an available text model with the Web Search workflow you need. For supplied documents, check file input and enough context for the source. Compare the same question on citation quality, source relevance and factual accuracy before choosing a model for the larger task. |
| Files and images | Check file or image input support, attach your source and let it finish processing. Ask a specific question about its contents. |
| Coding and tools | Use a model with the inputs and function-calling support your coding client needs. Provide the relevant code, expected behavior and checks. |
| Creative media | Choose an image, video, music or other media model in the corresponding generator. For example, Veo is a video-model family. |
- File and PDF inputsLearn more
- Image generationLearn more
- Video generationLearn more
- Music generationLearn more
Find model IDs and capabilities for the API
The public GET /v1/models catalog lists current model IDs, aliases, supported inputs and tools, context capacity and output limits. Copy a returned ID or documented alias into your client configuration.
curl -sS https://gab.ai/v1/modelsUseful model catalog fields
| Detail | How to use it |
|---|---|
| id and aliases | Use a returned model ID or documented alias in your request. |
| capabilities | Check inputs and features such as file_input, image_input, web_search and function_calling. |
| context_window | The model’s context capacity; a null value means the catalog does not specify it. |
| max_output_tokens | The listed output limit; a null value means the catalog does not specify it. |
Context length and output limits
The context window is the model’s working capacity for a request, including conversation and source material. The output limit controls how much it can generate; reasoning can also use that budget. If an answer stops mid-sentence, ask for a shorter response or adjust the output budget in your API request. For Chat Completions, finish_reason: length indicates a limit was reached; read the answer as well as its completion status.
Use Auto or choose a named model
Auto chooses a model for supported Chat requests. Choose a named model when you want a particular style or capability, or want to compare the same task consistently. In the API, use model="auto" for routing or a current model ID for a named choice; the response model and routing information identify what ran. Media generators have their own model pickers.
Compare answers on a task you know
Start a separate chat for each model with the same source, request and constraints. Keep the replies open in separate tabs to compare them side by side. Compare factual accuracy, missing details, tone and whether the answer follows your instructions. Then ask for one specific revision and check what changed. For research, compare the evidence behind the claims as well as the writing. A useful comparison is tied to your task, rather than a single best-model ranking.
Free model access and Plus
You can start with Arya and eligible text models using finite signup credits. Other models may require Plus or purchased credits. Check the selected model in the picker; the plans page explains subscriptions and credit packs. Auto uses the models available to your account.
Chat and API costs
On token-priced models, longer conversations, more source material and longer answers can increase usage. Tool-assisted tasks can involve additional model calls. In Chat, use the picker’s current cost information and your completed Usage History.
API requests require active Plus or Plus Premium and applicable credits, including requests to Arya; included Arya Chat use is separate. The catalog’s credit_cost field describes model pricing settings, not a fixed price for every API request.
Sources, search and provider context
For current research, enable Web Search and open the cited pages to check that they support the answer. When using your files, compare important details with the original.
A model name alone does not establish privacy. Choose based on who processes the request, what context you enable and the retention rules for your workflow. Relevant prompts, files and enabled context may be sent to the selected model provider through Gab’s provider account. Review the privacy reference before choosing what to share.
Resolve model or tool problems
If a model is unavailable, reopen the picker and check the selected entry and your account’s access. For an unsupported file or tool, choose a compatible model or adjust the input. If an answer is incomplete, review the output limit before retrying. Keep useful work and request a focused correction rather than restarting the whole task.
Put your chosen model to work.
Open Chat, choose a model and send a question, draft or source file.