Model Providers Reference
Look up provider identifiers, model types, credentials, and capability fields
Provider Identifiers
Current enum values: openai, openai_responses, anthropic, google, azure_openai, deepseek, moonshot, zhipu, qwen, baichuan, minimax, volcengine, siliconflow, xai, ollama, runway, pika, luma, kling, stability, midjourney, custom.
Model Types
chat, embedding, rerank, tts, stt, audio_generation, text_to_image, text_to_video.
Model Fields
| Field | Type | Default | Description |
|---|---|---|---|
name | string | Required | Display name, 1-100 characters |
provider | enum | Required | Provider identifier |
model_id | string | Required | Upstream model ID, 1-100 characters |
model_type | enum | Required | Model type |
provider_display_name | string/null | null | Provider name shown to users |
base_url | string/null | null | Custom API URL, at most 512 |
api_key | string/null | null | Can be empty for local providers; encrypted server-side |
context_length | integer/null | null | Must be >=1 |
max_output_tokens | integer/null | null | Must be >=1 |
input_price / output_price | decimal/null | null | Price per million Tokens, non-negative, up to 6 decimal places |
default_params | object/null | null | Default inference parameters |
capabilities | object/null | null | Capability flags such as vision, function calling, streaming |
is_enabled | boolean | true | Whether the model can be authorized/used |
is_default | boolean | false | Whether it is the default model of this type |
sort_order | integer | 0 | List sort order |
Connection tests return success, message, and optional latency_ms. When discovering models, credentials are only used for requests and are not saved automatically.
How is this guide?