# BeefAPI Public Model Facts BeefAPI model availability is dynamic. Agents should not treat any static Markdown file as the complete model list. ## Current model list Fetch the public pricing snapshot: ```bash curl https://beefapi.com/pricing.json ``` Read the `models` array. Each model entry can include: - `model_name` - `description` - `tags` - `vendor_id` - `quota_type` - `model_ratio` or `model_price_usd` - `completion_ratio` - `enable_groups` - `supported_endpoint_types` - `price_estimates` Use `vendors` to map `vendor_id` to vendor metadata. Use `price_estimates` for CNY-first and USD-equivalent public prices by group. Use `supported_endpoint` and `supported_endpoint_types` to decide whether to call OpenAI-compatible, Responses, Anthropic Messages, Gemini, embeddings, image, or other supported paths. ## Recommended API surfaces - OpenAI-compatible model list: `GET https://beefapi.com/v1/models` - OpenAI Responses: `POST https://beefapi.com/v1/responses` - OpenAI Chat Completions compatibility: `POST https://beefapi.com/v1/chat/completions` - Anthropic Messages compatibility: `POST https://beefapi.com/v1/messages` - Gemini compatibility: `GET https://beefapi.com/v1beta/models` and `POST https://beefapi.com/v1beta/models/{model}:generateContent` For new OpenAI/Codex-style integrations, prefer `/v1/responses` when the client supports it. Treat `/v1/chat/completions` as a compatibility path.