Analytics preference

We only load Google Analytics after you opt in. You can change this choice at any time from Cookie settings.

DataTable

The DataTable tab lets you create and manage structured data tables directly in Heym.

The DataTable tab lets you create and manage structured data tables directly in Heym. Tables can be used in workflows via the DataTable node for programmatic CRUD operations.

Creating a Table

  1. Navigate to the DataTable tab from the dashboard
  2. Click Create DataTable
  3. Enter a name and optional description
  4. Add columns with the column editor

Columns

Each column has:

PropertyDescription
NameColumn display name
Typestring, number, boolean, date, or json
RequiredWhether new rows must include this column
UniqueWhether values must be unique across rows
Default ValueAuto-filled when a row omits this column

Add, edit, or remove columns from the table detail view header.

Generate Columns with AI

Instead of adding columns by hand, you can describe the table in plain language (or paste sample JSON) and let an LLM propose the schema. Both entry points open the same dialog:

  • Generate with AI (DataTable list header) — creates a brand-new table from your description.
  • AI columns (table detail view, next to the column controls) — proposes new columns to append to the table you have open.

How it works

  1. Pick an LLM credential and model (defaults to your first LLM credential and its latest model).
  2. Type a description — for example "A table of books with title, author, page count, and whether I've read it" — or paste a JSON sample/schema.
  3. Click Generate. The model returns a suggested schema.
  4. Review the generated columns in an editable form: change the name, type, default, unique, and notEmpty (required) for each, delete unwanted columns, or Add column manually. For a new table you can also edit the AI-suggested name and description.
  5. Click Create Table (new table) or Add Columns (existing table) to save. Press Esc to close the dialog at any time.

Append-only for existing tables

When generating columns for an existing table, the model sees your current columns as context and proposes only new columns. Existing columns and their data are never modified or removed — generated columns are appended. Duplicate names (case-insensitive) are dropped automatically.

An LLM credential (OpenAI, Google, or Custom) is required. If you have none, add one in Settings first.

Managing Rows

  • Add row: Click the add button to create a new row
  • Inline edit: Double-click any cell to edit its value. Press Enter or click away to save, Escape to cancel
  • Delete row: Use the row action menu to remove a row
  • Pagination: Rows display 25 per page with page navigation

CSV Import/Export

Import

  1. Click Import CSV in the table detail view
  2. Select a CSV file to upload
  3. Preview the first 5 rows and column mapping
  4. Green columns are matched to existing table columns; yellow are unmatched
  5. Click Import to bulk-insert rows

Export

Click Export CSV to download all rows as a CSV file.

Sharing

Share tables with other users or teams:

  1. Open the share dialog from the table detail view
  2. Add users by email with read or write permission
  3. Add teams with read or write permission
  4. Read-only users can view data; write users can modify rows

Using in Workflows

The DataTable node connects tables to workflows. Operations include find, getAll, getById, insert, update, remove, and upsert. No credentials are required.

LLM Cost Table (system table)

A pinned System tables section appears above your user-created tables with a single entry: LLM Cost Table. This is a fixed-schema, per-user editable view of per-model pricing used by the Traces cost charts.

Where the data comes from

Heym keeps a global table of per-model pricing seeded from Helicone's public pricing endpoint (https://www.helicone.ai/api/llm-costs). The sync runs in the background the first time you open the table or Traces and is throttled to one fetch per 24 hours. Press Refresh to force an immediate re-sync. Helicone removes from the upstream list are not deleted locally, so historic cost lookups stay stable.

Schema (read-only columns)

ColumnMeaning
ProviderANTHROPIC, OPENAI, GOOGLE, … (informational, used by Helicone)
ModelModel identifier matched against trace.model
OpMatch operator: equals, startsWith, or includes
Input $/1MUSD per 1 million prompt tokens
Output $/1MUSD per 1 million completion tokens
Sourcehelicone, seed, or user

Customizing pricing

  • Edit a row to override its prices. Your override is per-user and is marked with a Customized badge; future Helicone syncs leave it alone.
  • Reset to default removes the override and restores the global value for that row.
  • Add Custom Model opens a dialog to create a user-only row (for example, a private fine-tuned model Helicone does not list). Custom rows are marked User added and deleting them removes them entirely.

Matching against traces

When the Traces dashboard computes cost it matches each trace.model against:

  1. Your overrides (exact match on model name) first
  2. Otherwise the global table — equals exact match takes priority over startsWith/includes; ties within the same operator are resolved by longest matching model prefix/substring

If no row matches, the model is listed under the Unpriced models notice on the Traces page; its cost contribution is zero until you add a custom row or override one.