Implementation:Helicone Helicone ModelConfig AuthorMetadata

Knowledge Sources	Helicone
Domains	Model Registry, Type System, TypeScript
Last Updated	2026-02-14 00:00 GMT

Overview

Concrete TypeScript type definitions for model specifications and author metadata, provided by the packages/cost/models/types.ts module.

Description

The ModelConfig interface captures a model's intrinsic properties: its display name, author, description, context length, maximum output tokens, creation date, input/output modality, tokenizer family, and an optional pinned version reference. It is used in per-model model.ts files within the packages/cost/models/authors/ directory hierarchy.

The AuthorMetadata interface describes an author organization: model count, support status, and optional fields for name, slug, description, website, and API URL.

Supporting types include AuthorName (a union derived from the AUTHORS const array plus "passthrough"), InputModality and OutputModality (unions of "text" | "image" | "audio" | "video"), Modality (an object with input and output arrays), and Tokenizer (a union of all supported tokenizer families).

The currently recognized authors are: anthropic, openai, google, meta-llama, mistral, amazon, microsoft, nvidia, deepseek, qwen, xai, moonshotai, perplexity, alibaba, zai, and baidu.

Usage

Use these types when:

• Creating a new model.ts file for a new model under packages/cost/models/authors/
• Building UI components that display model specifications
• Filtering or grouping models by author, modality, or tokenizer

Code Reference

Source Location

• Repository: Helicone
• File: packages/cost/models/types.ts (lines 3-11, 34-59, 154-164)

Signature

export interface AuthorMetadata {
  modelCount: number;
  supported: boolean;
  name?: string;
  slug?: string;
  description?: string;
  website?: string;
  apiUrl?: string;
}

export interface ModelConfig {
  name: string;
  author: AuthorName;
  description: string;
  contextLength: number;
  maxOutputTokens: number;
  created: string;
  modality: Modality;
  tokenizer: Tokenizer;
  pinnedVersionOfModel?: string;
}

export type AuthorName = (typeof AUTHORS)[number] | "passthrough";

export type InputModality = "text" | "image" | "audio" | "video";
export type OutputModality = "text" | "image" | "audio" | "video";

export interface Modality {
  inputs: InputModality[];
  outputs: OutputModality[];
}

export type Tokenizer =
  | "Claude" | "GPT" | "Llama" | "Llama3" | "Llama4"
  | "Gemini" | "Mistral" | "MoonshotAI" | "Qwen"
  | "DeepSeek" | "Cohere" | "Grok" | "Tekken" | "Zai" | "Baidu";

Import

import type { ModelConfig, AuthorMetadata, AuthorName, Modality, Tokenizer } from "packages/cost/models/types";

I/O Contract

Inputs

Name	Type	Required	Description
name	string	Yes	Display name of the model (e.g., "Anthropic: Claude Opus 4")
author	AuthorName	Yes	Author identifier from the AUTHORS const array
description	string	Yes	Human-readable description of the model's capabilities
contextLength	number	Yes	Maximum context window size in tokens
maxOutputTokens	number	Yes	Maximum number of tokens the model can generate per response
created	string	Yes	ISO 8601 creation date string
modality	Modality	Yes	Supported input and output modalities
tokenizer	Tokenizer	Yes	Tokenizer family used by the model
pinnedVersionOfModel	string	No	If this model is a version-pinned alias, the canonical model name it points to

Outputs

Name	Type	Description
ModelConfig	TypeScript interface	Type-checked model specification object
AuthorMetadata	TypeScript interface	Type-checked author metadata object

Usage Examples

Basic Usage

import type { ModelConfig } from "packages/cost/models/types";

// Define a new model's intrinsic properties
export const models = {
  "claude-opus-4": {
    name: "Anthropic: Claude Opus 4",
    author: "anthropic",
    description:
      "Our previous flagship model with very high intelligence and capability...",
    contextLength: 200000,
    maxOutputTokens: 32000,
    created: "2025-05-14T00:00:00.000Z",
    modality: { inputs: ["text", "image"], outputs: ["text"] },
    tokenizer: "Claude",
  },
} satisfies Record<string, ModelConfig>;

Related Pages

Implements Principle

• Principle:Helicone_Helicone_Model_Author_Metadata