Get started
Models
Forge exposes models from every major provider behind a single OpenAI-compatible API. Models are addressed with a vendor prefix.
Naming convention
texttext
<vendor>/<model-id> # examples anthropic/claude-4.5-sonnet openai/gpt-5 google/gemini-3-pro deepseek/deepseek-r2
Discovery
List every model your key can call via the metadata endpoint, or browse the catalog in the console.
bashbash
curl https://api.cytokenhub.com/v1/meta/model/list \ -H "Authorization: Bearer $FORGE_KEY"
Pricing
Token prices and context windows are returned by the metadata endpoint and shown on the Pricing page. Prices match upstream provider rates exactly — Forge marks up nothing on managed models.
Capabilities
- chat — every model supports
/v1/chat/completions - streaming — SSE on all chat models
- vision — image inputs on multimodal models (see
features: ["vision"]) - tools — function calling on OpenAI / Anthropic / Google flagship models
- json mode — structured outputs on supported models
- embeddings, images, audio — see API reference