Skip to main content
Screenshot 2026 06 01 At 11 34 48 AM

​
View Options

Toggle between two display modes to suit your preference:
  • Card view - Visual grid layout showing models as cards
  • Table view - Compact tabular format for detailed information

​
Base model

When deploying directly on 4MINDS, your custom model runs on gpt-oss-120b — no in-platform base-model picker.
Direct base-model selection has been deprecated. To use other foundation models, connect an external provider — see Creating a Model in Advanced Mode below for details.

​
Creating a Model in Advanced Mode

  1. Step 1 of 3 — Model Settings. Click the ‘Create Model’ button and select ‘Advanced Mode’ to begin creating your model. Configure the basics for your model, then click ‘Next’:
  • Name (required) — Enter a name for your model.
  • Description (required) — Briefly describe what the model is for.
  • Use Case (required) — Select the use case that best matches your model’s purpose.
  • Persona (optional) — Apply a persona to shape the model’s tone. See Personas.
  • Category (optional) — Tag the model with a category for organization.
  • Base model / External provider — By default, the model runs on gpt-oss-120b. To use a different foundation model, select an externally connected provider (e.g. Amazon Bedrock, Google Vertex AI, Amazon SageMaker, or Microsoft Foundry) — see Add Integrations & Data Sources for setup.
Screen Shot2025 11 01at7 40 17PM Pn

​

  1. Step 2 of 3 — Upload data. Add training data to your model. Select a data source using the tab selector at the top, or pick a pre-existing dataset from the Or use an existing dataset dropdown at the bottom.
Available data source tabs:
  • Upload (default) — Select local files using the Choose a File button. Multiple files can be uploaded at once. Supported formats include:
    • Data: CSV, TSV, PARQUET, JSON, JSONL
    • Documents: PDF, DOCX, MD, TXT
    • Code: PY, JS, TS, SQL, and many others
    • Archives: ZIP Maximum file size is 2,000 MB, with a total upload cap of 2,000 MB.
  • Integrations — Pull data from a connected integration (e.g. Hugging Face, S3, Google Drive).
  • URL — Import data from a public web URL.
Click ‘Add Files’ to attach the selected files to the model. Screen Shot2025 11 01at7 48 02PM Pn
  • Click ‘Next’ to proceed to the next step.
  1. Step 3 of 3 — Review. Verify your configuration before model creation begins. The review screen shows a read-only summary of the choices made in the previous steps in a two-column label/value layout:
FieldDescription
NameThe model’s display name (e.g. “Forecasting for Oil and Gas”)
DescriptionBrief summary of the model’s purpose
Base ModelThe underlying foundation model (e.g. GPT-OSS-120B)
Use CaseThe intended application domain (e.g. “Finance”)
PersonaThe assigned persona, or None if not configured
CategoryThe classification category for the model (e.g. “Finance”)
DatasetThe training dataset selected in Step 2
Review each field for accuracy. To correct a value, navigate back to the relevant step. When everything looks correct, click ‘Create Model‘ to finalize your model creation.