Catalog & Models

Deva has two catalogs: the API Catalog (agent-payable tools/APIs) and Models (LLMs). Both are browsable in the dashboard.

API Catalog

The API Catalog lists live, agent-payable resources — "grouped by the categories agents use most."

• Search by name, provider, or description.
• Sort — Most used (default), Name A–Z, or Provider A–Z.
• Filter by category — Social, Music & Audio, Video, E-commerce, Search & SEO, Finance & Crypto, Travel, Knowledge & Data.
• Get API key (top right) → Keys.

Each resource card shows its burn rate, sync/async kind, and endpoint count. Select Open in Playground to try it (this preselects it via ?api={slug}). Expand Endpoints to see the inspect / estimate / run / poll paths and which need a key. A Published burn-rate catalog table lists live per-resource rates.

→ Call these via the resource rail.

Models

The Models page lets you "compare every LLM model available through your Agent One Key by provider, capability, context window, and credit price."

• Header stats: Models, Providers, Recommended, Largest context.
• Search and Sort (Recommended, price, output tokens, context, name).
• Filter by Category, Provider, Capabilities (Tool calling, Reasoning, Vision, Structured output, Streaming), Min context length, Max credit price.
• Get an API key → Keys.

Model detail

Each model page has a Copy model id button, context/output/burn-rate stats, a capabilities list, related models, and a Quick start (#quickstart) with drop-in snippets — cURL, JavaScript, and SDK (@deva-me/agent-sdk) — against the OpenAI-compatible endpoint. The sidebar notes: "$1 equals 1 credit."

→ Calling models

Resources vs models

The API Catalog is payable tools (called via the resource rail); Models are LLMs (called via /v1/chat/completions). → Resources vs models

Next steps

• Playground — test any catalog API
• The resource rail · Calling models